Summer heat deaths in 854 European cities more than tripled due to climate change
[s.n]
Auteur moral
Auteur secondaire
Résumé
"Extreme heat is the deadliest type of weather and officially reported heat deaths remain significantly underestimated. We use established peer-reviewed methodology to estimate the total number of heat-related deaths in 854 cities during the recent heatwave and tocalculate the proportion of these deaths that can be attributed to climate change. In total, we estimate climate change driven changes to the temperatures to have caused 16,469 additional excess deaths (95% empirical Confidence Intervals: 15,013 to 17,864) across the 854 cities, accounting for almost 70% of the estimated summer deaths."
Editeur
Grantham Institute - Climate Change and the Environment
Descripteur Urbamet
Descripteur écoplanete
changement climatique
; risque sanitaire
Thème
Économie - Société
; Énergie - Climat
; Risques
; Ressources - Nuisances
; Santé
Texte intégral
Institute reports and analytical notes
Summer heat deaths in
854 European cities
more than tripled due to
climate change
2025
2 Grantham Institute Summer heat deaths in 854 European
cities more than tripled due to climate
change
Contents
INTRODUCTION 4
TRENDS IN SUMMER TEMPERATURES OVER EUROPE 8
HEALTH IMPACTS OF CHANGING SUMMER TEMPERATURES 13
DATA & METHODS 21
APPENDICES 28
3 Grantham Institute Summer heat deaths in 854 European
cities more than tripled due to climate
change
Key points
? Extreme heat is the deadliest type of weather and officially reported heat deaths remain
significantly underestimated. We use established peer-reviewed methodology to estimate
the total number of heat-related deaths in 854 cities during the recent heatwave and to
calculate the proportion of these deaths that can be attributed to climate change. In total,
we estimate climate change driven changes to the temperatures to have caused 16,469
additional excess deaths (95% empirical Confidence Intervals: 15,013 to 17,864) across the
854 cities, accounting for almost 70% of the estimated summer deaths.
? June-August 2025 was the fourth warmest summer season on record, 0.9°C above the
1990-2020 mean. This means a seriously heightened risk of death for vulnerable people,
including those over 65 and with preexisting medical conditions, as well as increased risk of
overheating of indoor environments. All of this comes with an increasing high demand for
health services and increased power demand.
? We use both observations-based products and climate models to estimate that summer
temperatures across Europe are now 1.5 to 2.9°C than they would have been in a 1.3°C
cooler climate, without human-caused climate change caused primarily by the burning of
fossil fuels. However, this is probably a conservative estimate of the true warming
experienced in many cities, as climate models are known to underestimate warming in
Europe.
? Among 30 European capitals, Rome, Athens, and Bucharest had the highest estimated
excess mortality per population this summer. The largest relative proportions of heat-related
deaths were observed in Stockholm, Madrid, and Bratislava, with more than 85% of the
estimated summer deaths to be attributed to climate change.
? The study did not investigate the factors that influenced the deaths estimated in each city or
the current level of preparedness for mitigating the heat-related health risks. It is likely that
most deaths estimated here reflect the intensity of heat experienced by each city this
summer, but factors including population demographics, air pollution and adaptation to heat
are also known to play an important role.
? The study looks at 854 cities in the EU and affiliated countries. This covers only about 30%
of the population of Europe and so cannot be taken as indicative of the total number of
excess deaths. A range of limitations may have influenced the estimates, including climate
models that are known to underestimate the rate of warming, an event definition that does
not capture the most intense peaks in daily temperatures and exposure-response functions
that do not consider attenuation to heat in recent years.
? Cities are highly vulnerable to heatwaves because large amounts of concrete and asphalt
surfaces trap and hold heat, while transport and energy use generate even more,
intensifying dangerous urban temperatures. Expanding green and blue spaces are known to
decrease this urban heat island effect and provide crucial cooler spaces that can be lifelines
for people in heatwaves, particularly lower socioeconomic groups that live in denser housing
and are less likely to have air conditioning. Urbanisation is an increasing trend in Europe with
70% of people living in cities and more than 80% expected by 2050. The converging trends
4 Grantham Institute Summer heat deaths in 854 European
cities more than tripled due to climate
change
of urbanisation, ageing population and climate change drive vulnerability and increase the
risk of reaching limits to adaptation.
? Current policies in place around the world are projected to result in about 2.7°C of warming
above pre-industrial levels by 2100, which would result in many more heat-related deaths
and impacts globally. This highlights yet again the urgent need to rapidly transition away
from fossil fuels to secure a liveable future.
Introduction
Summer 2025 has been ?roasting hot? across Europe, with back-to-back heatwaves affecting both
humans and the economy (Guglielmi, 2025). Peaking in mid-June to early July across different
regions, Western Europe as a whole experienced its highest average temperatures for this period in
decades, and the hottest June on record (ibid.). Temperatures rose above 40°C, and up to 46°C in
Spain and Portugal, due to a persistent high-pressure system, also known as a heat dome, causing it
to stay hotter for longer (ibid.). Fennoscandia also experienced an exceptionally persistent
heatwave in July, including the longest streak of daily maximum temperatures reaching 25°C ever
observed in Finnish Lapland - 26 consecutive days (Barnes et al., 2025). Southeast Europe faced
heatwaves in July and a national record temperature of 50.5°C was recorded in Türkiye
(Copernicus, 2025).
In August, the entire Mediterranean region of Europe, especially the Iberian Peninsula, was affected
by another intense heat wave, reinforced by both a heat dome and a heat plume, the rising of hot air
masses from North Africa and the Iberian Peninsula (Le Monde, 2025). Spain recorded its most
intense heat wave ever, exceeding the 2022 event with a temperature anomaly of 4.6°C, according
to provisional data from Spanish State Meteorological Agency AEMET (Euronews, 2025). From 8 to
17 August, Spain experienced the hottest ten-day period ever recorded since 1950 at least
(Euronews, 2025). Unprecedented heat levels were also recorded from the southwest to the
center-east of France, as well as near the Mediterranean (Météo France, 2025).
At the European scale, June 2025 was the fifth-warmest June on record (Copernicus, 2025a), while
July 2025 ranked as the fourth-warmest July on record (Copernicus, 2025b), and August the third
warmest August (Copernicus, 2025c). Overall, it was the fourth warmest summer season on record,
0.9°C above the 1990-2020 mean (ibid.). When looking at the summer of 2025 as a whole and
individual countries, it was the warmest on record for the UK, with a mean temperature of 16.10°C,
surpassing the previous record of 15.76°C set in 2018 (Met Office, 2025). France had its third
hottest summer on record, behind the summers of 2003 and 2022 (Météo France, 2025). However,
parts of eastern Europe, including the Baltic states, Belarus and Poland, were slightly cooler than
the 1990-2020 average (Figure 1).
mailto:https://www.nature.com/articles/d41586-025-02553-3
https://www.nature.com/articles/d41586-025-02553-3
https://www.nature.com/articles/d41586-025-02553-3
https://spiral.imperial.ac.uk/entities/publication/905c62f2-4268-45b7-8310-2de66a26a9e8
https://climate.copernicus.eu/surface-air-temperature-july-2025
https://www.lemonde.fr/en/environment/article/2025/08/21/august-heat-wave-deemed-a-rare-event-by-meteo-france_6744593_114.html
https://www.euronews.com/green/2025/08/26/summer-2025-ranks-among-spains-hottest-ever-after-historic-august-heatwave
https://www.euronews.com/green/2025/08/26/summer-2025-ranks-among-spains-hottest-ever-after-historic-august-heatwave
https://meteofrance.com/actualites-et-dossiers/actualites/vague-de-chaleur-du-8-au-18-aout-2025-retour-sur-un-episode#:%7E:text=Les%20vagues%20de%20chaleur%20en%20ao%C3%BBt%20en%20France&text=Plus%20r%C3%A9cemment,%20en%202023,%20une,et%20pr%C3%A8s%20de%20la%20M%C3%A9diterran%C3%A9e.
https://climate.copernicus.eu/surface-air-temperature-june-2025
https://climate.copernicus.eu/surface-air-temperature-july-2025
https://climate.copernicus.eu/surface-air-temperature-august-2025
https://climate.copernicus.eu/surface-air-temperature-august-2025
https://www.metoffice.gov.uk/about-us/news-and-media/media-centre/weather-and-climate-news/2025/summer-2025-is-the-warmest-on-record-for-the-uk
https://meteofrance.com/actualites-et-dossiers/actualites/lete-2025-au-3-rang-des-etes-les-plus-chauds
5 Grantham Institute Summer heat deaths in 854 European
cities more than tripled due to climate
change
Figure 1: Mean temperature anomaly in June-August 2025 compared to 1990-2020 climatology
(ERA5). The dark blue line indicates the regions considered within this study, which is bounded by
the domain 11W, 35E; 33N, 66N.
While the temperatures reached across the continent in 2025 were unusually warm, temperatures
across Europe are known to be increasing more rapidly than the rest of the world (Copernicus,
2024), and there is an extensive body of literature attributing this with high confidence to human
influence on the climate (Seneviratne et al., 2021; Quilcaille et el., 2025).
Heat, extreme heat and heatwaves affect human health and well-being. The European Environment
Agency (EEA) describes heat as ?the largest and most urgent climate hazard for human health?
(EEA, 2024). In the past, Europe has recorded thousands of excess heat-related deaths each year
(Masselot, et. al., 2023, Ballester et al., 2023). During the summer of 2022, more than 60,000
people across Europe died as a result of extreme heat (Ballester et al., 2023). An impact attribution
study found that 56% of these excess deaths could be attributed to human-induced climate change
(Beck et al., 2024). Even in the following summer, which was less intensely hot, over 47,000 heat-
related deaths were recorded (Gallo et al., 2024). A global study recently showed that 178,000
deaths can be linked to heatwaves in 2023 and that more than half (54.29%) of these deaths are
attributable to climate change (Hundessa et al., 2025). It also found that the highest mortality rates
and excess death ratios occurred in Southern, Eastern and Western Europe, amounting to a total of
66,443 for the continent (Hundessa et al., 2025).
Among other factors, population age structure and socio-economic drivers determine vulnerability
to high temperatures (Falchetta et al., 2024). Europe's population is projected to age significantly
https://climate.copernicus.eu/esotc/2024
https://climate.copernicus.eu/esotc/2024
https://www.ipcc.ch/report/ar6/wg1/downloads/report/IPCC_AR6_WGI_Chapter11.pdf
https://www.nature.com/articles/s41586-025-09450-9
https://www.eea.europa.eu/en/analysis/publications/european-climate-risk-assessment
https://www.sciencedirect.com/science/article/pii/S2542519623000232?via%3Dihub
https://www.nature.com/articles/s41591-023-02419-z
https://www.nature.com/articles/s41591-023-02419-z
https://www.nature.com/articles/s41612-024-00783-2
https://www.nature.com/articles/s41591-024-03186-1?fromPaywallRec=false
https://www.sciencedirect.com/science/article/pii/S2666675825003133?via%3Dihub
https://www.sciencedirect.com/science/article/pii/S2666675825003133?via%3Dihub
https://www.nature.com/articles/s41467-024-47197-5
6 Grantham Institute Summer heat deaths in 854 European
cities more than tripled due to climate
change
(Eurostat, 2023), and by 2050, a substantial share of Europe?s population is expected to be exposed
to extreme heat (Falchetta et al., 2024). Under a 2°C global warming scenario, approximately 163
million Europeans could face unprecedented summer temperatures, nearly twice the number
currently affected (King et, al., 2018; Garcia-Leon et al., 2024). While there is quantitative evidence
that adaptation to extreme heat is taking place to a certain extent in some places (Vicedo-Cabrera
et al., 2018, Stuart-Smith et al., 2025), large adaptation gaps still prevail and future impacts from
extreme heat can still be avoided.
Beyond causing deaths, heat increases hospitalisations and worsens chronic illnesses, especially
for older adults and low-income communities (World Bank, 2025). Livelihoods and economic
activities are threatened as jobs in informal services, construction, transport and tourism are
particularly hit (World Bank, 2025).
Adaptation to heat
Cities are hotspots of heat risk due to the urban heat island effect, rapidly ageing populations, and
urban growth pressures; urbanisation is an increasing trend in Europe with 70% of people living in
cities and more than 80% expected by 2050 (World Bank, 2025; European Commission, 2025a). A
wide portfolio of adaptation measures is required to reduce impacts: from early-warning systems
and timely advice to the public to the enhancement of housing and urban planning (such as
expanding green spaces), reinforced healthcare and social services and the adjustment of working
conditions and activities during periods of extreme heat (EEA, 2025). While impacts of heat on
human health are recognised in a large share of national adaptation and health policies and
strategies, the level of policy preparedness to heat for Europe is assessed as ?medium? by the
European Climate Risk Assessment, due to aspects of social justice often missing in planning
instruments (EEA, 2024). Heat-health action plans are a key tool to reduce deaths during extreme
heat, as they aim to assign responsibilities in the event of an emergency and plan both short- and
long-term measures to reduce risks (EEA, 2025). Current mitigation policies in place around the
world are projected to result in about 2.7°C warming above pre-industrial levels by 2100 (Climate
Action Tracker, 2024), which would result in many more heat-related deaths and impacts globally.
Research has shown that both stringent mitigation and adaptation must be implemented in parallel,
as it is becoming increasingly challenging to adapt (Masselot, 2025) and limits to adaptation are
being reached (IPCC, 2022).
In addition, as the results presented here and large literature on vulnerability to heat shows, older
people are particularly vulnerable. Even if the frequency or intensity of climate hazards were to
remain constant due to effective mitigation, collective risks from extreme weather events have the
potential to increase significantly in the future due to an ageing population (Harrington and Otto,
2023; European Commission, 2025b). While policies promoting sustainable development inherently
support communities in managing the risks associated with extreme weather over the coming
decades, additional targeted strategies will be necessary to address the unique challenges posed
by rapidly growing older populations projected under different development pathways. Failing to
account for these demographic risks could further complicate efforts to achieve well-being and
resilience in a warming climate.
https://ec.europa.eu/eurostat/statistics-explained/index.php?title=Population_projections_in_the_EU
https://www.nature.com/articles/s41467-024-47197-5
https://www.nature.com/articles/s41558-018-0191-0
https://www.sciencedirect.com/science/article/pii/S2468266724001798?via%3Dihub
https://www.sciencedirect.com/science/article/pii/S0160412017310346
https://www.sciencedirect.com/science/article/pii/S0160412017310346
https://link.springer.com/article/10.1007/s10584-025-04011-5
https://www.worldbank.org/en/region/eca/publication/unlivable-how-cities-in-europe-and-central-asia-can-survive-and-thrive-in-a-hotter-future
https://www.worldbank.org/en/region/eca/publication/unlivable-how-cities-in-europe-and-central-asia-can-survive-and-thrive-in-a-hotter-future
https://www.worldbank.org/en/region/eca/publication/unlivable-how-cities-in-europe-and-central-asia-can-survive-and-thrive-in-a-hotter-future
https://knowledge4policy.ec.europa.eu/foresight/topic/continuing-urbanisation/developments-and-forecasts-on-continuing-urbanisation_en#:%7E:text=Urbanisation%20in%20Europe,population%20between%202015%20and%202050
https://climate-adapt.eea.europa.eu/en/observatory/evidence/health-effects/heat-and-health
https://www.eea.europa.eu/en/analysis/publications/european-climate-risk-assessment
https://climate-adapt.eea.europa.eu/en/observatory/evidence/health-effects/heat-and-health
https://climateactiontracker.org/global/cat-thermometer/
https://climateactiontracker.org/global/cat-thermometer/
https://www.nature.com/articles/s41591-024-03452-2
https://www.ipcc.ch/report/ar6/wg2/downloads/report/IPCC_AR6_WGII_SummaryForPolicymakers.pdf
https://www.nature.com/articles/s41612-023-00398-z
https://www.nature.com/articles/s41612-023-00398-z
https://ec.europa.eu/eurostat/statistics-explained/index.php?title=Population_structure_and_ageing
7 Grantham Institute Summer heat deaths in 854 European
cities more than tripled due to climate
change
Study aims & scope
This study first uses an established extreme event attribution methodology to analyse changes in
monthly mean temperatures during the summer months of 2025 in 854 cities across Europe
(Figure 2), and the extent to which those changes can be attributed to human-caused warming. We
use this attributable change in intensity to estimate how much cooler the daily mean temperatures
would have been throughout the summer months, if human activities had not warmed the planet by
1.3°C above preindustrial levels. We use established exposure-response relationships (Masselot et
al., 2023) to estimate the excess mortality attributed to heat given the observed temperatures and
also under the counterfactual scenario of a 1.3°C cooler world. By comparing these estimates, we
calculate how many more people are expected to have died in each city during the summer of 2025
due to the excess heat caused by anthropogenic warming. We also report the expected proportion
of heat-related deaths attributed to human-induced climate change. The 854 cities represent all
cities in the EU and affiliated countries (including Switzerland and the UK) with more than 50,000
inhabitants, totalling an estimated 30% of the total population of Europe.
Figure 2: Mean temperature anomaly June-August 2025 compared to 1990-2020 climatology in
each of the 854 cities included in the study (ERA5).
At the time of writing, the actual number of deaths reported during the study period was not yet
available; therefore, the numbers reported here should be interpreted as estimates of expected
heat related mortality rather than actual outcomes. Briefly, to estimate the expected mortality, we
use all available historical mortality data for each city to calculate a baseline mortality rate. This rate
is assumed to remain constant throughout the year and is projected to apply to the year 2025.
https://www.thelancet.com/journals/lanplh/article/PIIS2542-5196(23)00023-2/fulltext
https://www.thelancet.com/journals/lanplh/article/PIIS2542-5196(23)00023-2/fulltext
8 Grantham Institute Summer heat deaths in 854 European
cities more than tripled due to climate
change
1 Trends in summer temperatures over Europe
1.1 Summer 2025
Summer temperatures in 2025 were around 0.9°C above the 1990-2020 mean over the whole of
Europe (Copernicus, 2025c), although there was substantial variability within the region, with
southern and eastern Europe experiencing much higher temperatures than northern and western
Europe (Figure 3a). With the exception of Estonia, Latvia and Lithuania, all countries within the
study region experienced a warmer than average summer (Figure 3b), with much of western and
southern Europe experiencing temperatures more than 1°C warmer than the 1990-2020 mean.
Figure 3: (a) mean summer temperatures and (b) mean summer temperature anomaly with respect
to 1990-2020 climatology, for all countries within the scope of the study. Mean temperatures are
averages over daily mean temperatures from ERA5 reanalysis, encompassing all land grid cells
within the domain 11W, 35E; 33N, 66N (shown in Figure 1). Black lines divide countries into Northern,
Western, Eastern and Southern regions.
Figure 4 shows the daily mean temperature anomalies in each country from June-August 2025 with
respect to the 1990-2020 daily climatology, highlighting the different spatial patterns of warm
periods across the continent. Much of continental Europe was unusually warm throughout June,
culminating in a heatwave at the end of the month during which much of western and southern
Europe was more than two degrees warmer than usual for the time of year (Figure 4, box 1). Heat
health warnings were issued in many cities and the hot, dry conditions led to a reported 300
hospitalisations due to the heat, with eight fatalities reported while the event was still ongoing
(ERCC, 2025), as well as creating extremely fire-prone conditions (Clarke et al., 2025).
https://climate.copernicus.eu/surface-air-temperature-august-2025
https://erccportal.jrc.ec.europa.eu/ECHO-Products/Echo-Flash#/daily-flash-archive/5375
https://www.imperial.ac.uk/grantham/publications/all-publications/climate-change-tripled-heat-related-deaths-in-early-summer-european-heatwave.php
9 Grantham Institute Summer heat deaths in 854 European
cities more than tripled due to climate
change
While much of Europe was slightly cooler throughout July, Scandinavia and the Baltic countries
experienced a lengthy heatwave from mid to late July during which multiple heat records were set
across Fennoscandia (Figure 4, box 2). Increased heat-related hospitalisations were reported across
the region, and numerous wildfires broke out, with the resulting smoke impacting air quality (Barnes
et al., 2025). It is important to note that the gridded data used to estimate the national mean
temperatures was clipped at 66N in order to ensure that the temperatures reported are more or
less representative of those experienced in the cities for which mortality data was available.
Moreover, since the highest temperatures and temperature anomalies recorded during this
heatwave occurred further north than this, the figures reported here should not be taken as
representative of the temperatures over these countries as a whole, which are likely somewhat
higher.
A record-breaking week-long heatwave in late July (Figure 4, box 3) in southeast Europe was linked
to devastating wildfires in Greece, Cyprus, Türkiye and the rest of the Balkans (Keeping et al.,
2025a; EFFIS, 2025). We note that the six Balkan countries outside of the EU are not represented in
this study; however, as Figure 1 shows, they also experienced a very warm summer, with the
average summer temperature across Albania, Bosnia and Herzegovina, Kosovo, Montenegro, North
Macedonia and Serbia more than 1.7°C warmer than the 1990-2020 mean.
In mid-August, a second pan-European heatwave again brought temperatures more than 2°C above
the seasonal mean for nearly a week (Figure 4, box 4). In Spain and Portugal, where the heat
persisted for several weeks, this was one of the most intense heatwaves on record, again coinciding
with record-breaking levels of wildfire activity (Keeping et al., 2025b).
For the four periods of extreme heat highlighted in Figure 4 (dark blue boxes), unusually, health
impacts were reported at the time of the events. However, health impacts from heatwaves are
often not reported while the event is ongoing; and even less severe heatwaves pose significant
risks. In the next Section 1.2, we evaluate the extent to which the continent has been warmed by
human-caused climate change. Section 2 then assesses the expected risks to human health
associated not only with the most intense heatwave events but over the whole summer period.
https://spiral.imperial.ac.uk/entities/publication/905c62f2-4268-45b7-8310-2de66a26a9e8
https://spiral.imperial.ac.uk/entities/publication/905c62f2-4268-45b7-8310-2de66a26a9e8
https://hdl.handle.net/10044/1/123302
https://hdl.handle.net/10044/1/123302
https://doi.org/10.25560/123547
10 Grantham Institute Summer heat deaths in 854 European
cities more than tripled due to climate
change
Figure 4: Daily mean temperature anomaly with respect to 1990-2020 daily climatology, averaged
across all grid cells within the country bounded by the domain 11W, 35E; 33N, 66N (ERA5). Dark
blue boxes highlight periods of particularly dangerous heat that have previously been the subject of
attribution studies (see text for details).
1.2 Changes in monthly mean temperatures
To understand the extent to which summer temperatures across Europe have been affected by
human-caused climate change, we use an established probabilistic attribution protocol (Philip et al.,
2020). The long-term trend in global mean temperatures is known to be fully driven by human
activity, leaving the climate in 2025 an estimated 1.3°C warmer than it would have been without this
anthropogenic warming. We use a statistical model to represent the relationship between monthly
mean temperatures in each location and global mean temperatures, and so to describe the effect of
this human-caused warming on local temperatures. Historic trends in temperatures are assessed in
two observation-based gridded datasets, along with trends in the output of 38 climate models from
https://doi.org/10.5194/ascmo-6-177-2020
https://doi.org/10.5194/ascmo-6-177-2020
11 Grantham Institute Summer heat deaths in 854 European
cities more than tripled due to climate
change
the most recent CMIP6 experiments (Table A2). For each of these sources we estimate the local
change in temperature associated with 1.3°C of warming, and finally combine the results into a
single estimate known as the synthesis (Otto et al., 2024). This process is repeated for each of the
boreal summer months (June, July and August) and in each of the 854 cities. A more detailed
description of the datasets and methods used can be found in Section 3.1. Full uncertainty ranges
were also obtained for each city and month via bootstrapping, for use in the mortality attribution
analysis described in the next section, but are not presented in detail here; all ranges reported in the
text are ranges of estimated changes across all cities within each country or region.
The results presented here use the synthesised change in temperatures, taking into account both
historical trends and simulations from global climate models. This reduces the uncertainty in
observed trends, which can be high; however, observed warming trends across Europe in recent
decades are typically somewhat stronger than those simulated by climate models, so the
temperature increases reported here should be taken as a conservative estimate of the true
warming that people are experiencing across the continent. This is discussed in more detail in
section 3.3.2.
Figure 5 shows the spatial pattern of the synthesised changes in summer temperatures across
Europe, along with the mean change per month in across all of the cities in each country; maps of
the synthesised change in each month can be found in the Appendix in Figure A1, and a table
including the range of estimates within each country in Table 1.
Over Europe as a whole there is a northwest-southeast gradient, with central, southern and eastern
Europe warming substantially more than countries in the northern region. Warming also tends to be
stronger further inland, with the lowest estimated rates of warming recorded in Ireland, Cyprus,
Malta and Portugal, countries where all of the cities in the study are in coastal regions. On average
the 854 cities have warmed by an estimated 1.9°C in June (range of estimates: 0.8 to 3.3°C); 2.2°C
in July (0.7 to 3.5°C), and 2.3°C in August (0.9 to 3.6°C). In 825 of the 854 cities (97%) summer
temperatures have warmed by more than 1.5°C on average; in 569 of the 854 cities (67%), by more
than 2°C; and 132 cities (15%), by more than 2.5°C. These results are entirely consistent with the
findings of previous attribution studies across Europe, where temperatures are known to be
warming more rapidly than the global mean (Copernicus, 2024).
https://ascmo.copernicus.org/articles/10/159/2024/
https://climate.copernicus.eu/esotc/2024
12 Grantham Institute Summer heat deaths in 854 European
cities more than tripled due to climate
change
Table 1: Mean summer temperature and anomaly with respect to 1990-2020, averaged over all land
grid cells in ERA5 in the country and within each domain 11W, 35E; 33N, 66N (shown in Figure 1).
?Attributable change in intensity? is the average of the synthesised change in temperatures taking
into account trends in both observational data products (ERA5 and E-Obs) and climate models from
June-August over all cities and months, with the range of the estimated changes given in brackets.
Region Country # cities Mean JJA
temperature
(°C)
Mean JJA
anomaly
(°C)
Synthesised
change in
JJA
(°C)
Northern Norway 4 13.04 0.88 1.7 (1.4 - 2.2)
Sweden 14 14.92 0.63 1.8 (1.2 - 2.1)
Finland 9 15.65 0.51 1.9 (1.5 - 2.2)
Estonia 3 16.40 -0.08 2.0 (1.6 - 2.2)
Latvia 10 16.50 -0.26 2.0 (1.5 - 2.2)
Lithuania 6 16.86 -0.34 2.0 (1.5 - 2.3)
Denmark 4 16.65 0.47 1.6 (1.1 - 2.0)
United
Kingdom
135 16.10 1.49 1.8 (1.1 - 2.4)
Ireland 5 15.63 1.19 1.1 (0.7 - 1.5)
Western France 72 20.40 1.56 2.3 (1.4 - 3.3)
Belgium 15 18.70 1.38 2.1 (1.5 - 2.5)
Netherlands 47 18.39 1.10 2.0 (1.6 - 2.5)
Luxembourg 1 18.33 0.97 2.3 (2.2 - 2.5)
Switzerland 12 15.76 1.49 2.6 (2.0 - 3.2)
Germany 127 18.41 0.78 2.2 (1.4 - 2.8)
Austria 6 16.99 1.03 2.6 (1.9 - 2.9)
Eastern Poland 68 18.26 0.14 2.2 (1.5 - 3.0)
Czechia 18 18.23 0.48 2.4 (1.8 - 3.1)
Slovakia 8 19.01 0.65 2.6 (1.9 - 3.2)
Hungary 19 22.30 0.89 2.4 (1.8 - 3.0)
Romania 35 21.51 1.26 2.6 (1.8 - 3.5)
Bulgaria 18 23.13 1.72 2.4 (1.7 - 2.9)
Southern Greece 14 25.09 1.43 2.0 (1.1 - 2.6)
Cyprus 3 27.74 0.16 1.5 (1.4 - 1.6)
Croatia 7 22.75 1.22 2.5 (2.0 - 3.0)
Slovenia 2 20.47 0.97 2.7 (2.1 - 3.2)
Italy 87 22.87 1.31 2.4 (1.4 - 3.5)
Malta 1 25.84 1.34 1.6 (1.6 - 1.7)
Spain 90 24.11 1.87 2.2 (1.3 - 3.6)
Portugal 14 23.41 1.72 1.5 (0.9 - 2.2)
13 Grantham Institute Summer heat deaths in 854 European
cities more than tripled due to climate
change
Figure 5: (left) Map of synthesised change in summer temperatures averaged over June, July and
August and (right) panel plot showing the mean of the estimated changes in each country in each
month.
2 Health impacts of changing summer temperatures
In this section, we use established exposure-response relationships (Masselot et al., 2023) to
estimate the excess mortality given the observed temperatures (the ?factual? scenario) and also
under the counterfactual scenario of a 1.3°C cooler world. To calculate the exposure response
relationships, we used historical mortality data (1990-2019), and an established three-step
methodological framework that accounts for lagged effects and addresses data sparsity through
pooled estimates and extrapolation techniques (Masselot et al., 2023). The factual temperatures
are the daily mean temperatures in each city in the ERA5 reanalysis (Hersbach et al., 2020), and the
counterfactuals are calculated by subtracting the synthesised change in temperatures in each city
associated with 1.3°C of warming since the preindustrial period, as discussed in section 1.2. By
comparing these estimates, we calculate how many more people are expected to have died in each
city during the summer of 2025 due to the excess heat caused by anthropogenic warming.
2.1 Overall results
Overall, we found 24,404 (95% empirical Confidence Intervals: 21,968 to 26,806) excess deaths
due to heat during summer 2025, with 68% (64% to 71%) estimated to be attributable to human-
https://www.thelancet.com/journals/lanplh/article/PIIS2542-5196(23)00023-2/fulltext
https://www.thelancet.com/journals/lanplh/article/PIIS2542-5196(23)00023-2/fulltext
https://rmets.onlinelibrary.wiley.com/doi/full/10.1002/qj.3803
14 Grantham Institute Summer heat deaths in 854 European
cities more than tripled due to climate
change
induced climate change (Table 2). Older populations are affected the most, with more than 80% of
the excess heat-related deaths occurring among people older than 65 years.
Table 2: Population figures, median and 95% empirical confidence intervals for estimated excess
heat-related deaths presented as crude numbers and rates per 1 million population for deaths
attributable to heat, number of deaths attributable to human-induced climate change and
proportion of excess heat-related deaths that can be attributed to human-induced climate change.
Note that a minor rounding error occurs when summing the crude numbers by age group, resulting
in a slight difference from the overall total.
Age group Population Excess
deaths
Rate per 1
million
population
Attributable
to climate
change
Proportion
of excess
deaths due
to climate
change
Total 158,473,649 24404
(21968,
26806)
154
(139, 169)
16496
(15013, 17864)
0.68
(0.64, 0.71)
20-44 66,886,597 608
(434, 795)
9
(6, 12)
279
(208, 352)
0.46
(0.36, 0.57)
45-64 54,059,644 3058
(2557, 3593)
57
(47, 66)
1866
(1589, 2133)
0.61
(0.55, 0.67)
65-74 19,676,737 3738
(3279, 4171)
190
(167, 212)
2487
(2210, 2738)
0.67
(0.63, 0.7)
75-84 12,762,858 7019
(6305, 7726)
550
(494, 605)
4835
(4408, 5250)
0.69
(0.66, 0.72)
85+ 5,087,812 9959
(9012, 10964)
1957
(1771, 2155)
7028
(6397, 7657)
0.71
(0.67, 0.73)
2.2 Spatial variation
To allow comparison between cities across Europe, accounting for the different age distributions
among populations, we calculate the age-standardised excess mortality. Briefly, the excess deaths
were first converted into age-specific mortality rates by dividing by the corresponding population.
Standardised rates for each city are then weighted averages of these rates, using the 2013
European standard population as the weighing reference. The geographical distribution of the
median age-strandardised excess mortality and the median proportion of heat-related deaths
estimated to be attributed to human-induced climate change are shown in Figure 6. When
assessing the spatial distribution of the median age-standardised excess mortality in the 854 cities
in Europe, there is a north to south gradient, with cities in southern and southeast Europe affected
the most, where summer temperature anomalies were highest (Figure 2). Although the excess
mortality rates are lower in northern Europe, mainly because temperatures were lower, the
proportion of deaths attributable to climate change is higher. This spatial pattern closely reflects
the distribution of changes in Figure 5.
15 Grantham Institute Summer heat deaths in 854 European
cities more than tripled due to climate
change
When aggregating the available cities at the country level, we found the highest age-standardised
excess mortality due to heat in Croatia (382; 289 to 469), Greece (380; 312 to 442), and Bulgaria
(371; 267 to 477) (Table A1). The countries most affected by human-induced climate change, in
relative terms, were Sweden, Malta, and Slovakia, where 83% of the estimated heat-related deaths
were found to be attributable to climate change (Table A1).
When examining the 30 available European capitals, we estimated that Rome, Athens and
Bucharest had the highest standardised excess mortality per 1 million population (Table 3). In
absolute terms, we estimated 835 (712 to 938) heat-related deaths that could be attributable to
climate change in Rome, 630 (541 to 717) in Athens and 409 (310 to 511) in Paris. In relative terms,
the largest proportion of heat-related deaths due to climate change was observed in Stockholm
(97%; 91% to 98% ), Madrid (93%; 76% to 98%) and Bratislava (85%; 70% to 93%).
Figure 6: Maps of median standardised excess mortality per million people and median proportion
of mortality attributed to climate change by city.
Table 3: City specific population figures (based on the capitals), median and 95% empirical
confidence intervals for excess heat-related presented as crude numbers, age-standardised rates
(using the 2013 European standard population) per 1 million population, number of deaths
attributable to human-induced climate change and proportion of excess heat-related deaths that
can be attributed to human-induced climate change. The numbers reflect the heat-related impacts
during the entire summer of 2025.
16 Grantham Institute Summer heat deaths in 854 European
cities more than tripled due to climate
change
Capital Population Excess
deaths
Standardise
d excess
mortality per
million
Attributable
to climate
change
Proportion
of excess
deaths due
to climate
change
Paris 6,869,559 582
(433, 754)
95
(71, 123)
409
(310, 511)
0.70
(0.60, 0.73)
London 5,894,656 458
(302, 600)
114
(76, 149)
315
(215, 403)
0.69
(0.67, 0.71)
Madrid 2,871,466 423
(298, 574)
148
(104, 202)
387
(268, 497)
0.93
(0.76, 0.98)
Berlin 2,805,374 219
(174, 293)
82
(66, 110)
140
(115, 164)
0.66
(0.52, 0.69)
Athens 2,269,492 1093
(905, 1274)
470
(389, 548)
630
(541, 717)
0.58
(0.52, 0.63)
Rome 2,158,892 1280
(1135, 1432)
532
(472, 596)
835
(712, 938)
0.65
(0.59, 0.70)
Bucharest 1,665,377 472
(359, 609)
352
(269, 452)
360
(282, 442)
0.76
(0.69, 0.81)
Lisbon 1,425,616 84
(62, 121)
56
(41, 80)
50
(40, 67)
0.60
(0.55, 0.66)
Budapest 1,414,149 142
(85, 230)
106
(63, 173)
111
(71, 163)
0.79
(0.69, 0.86)
Warsaw 1,381,256 83
(51, 128)
60
(37, 93)
68
(43, 97)
0.82
(0.74, 0.87)
Vienna 1,315,748 111
(76, 160)
96
(66, 139)
86
(63, 112)
0.77
(0.67, 0.84)
Stockholm 1,142,160 31
(18, 45)
33
(18, 47)
30
(17, 42)
0.97
(0.91, 0.98)
Prague 990,604 46
(26, 73)
53
(31, 84)
36
(22, 54)
0.80
(0.71, 0.85)
Sofia 950,740 164
(106, 240)
214
(138, 311)
120
(81, 161)
0.72
(0.64, 0.80)
Dublin 893,653 10
(4, 17)
18
(7, 30)
8
(3, 13)
0.75
(0.73, 0.79)
Brussels 808,139 84
(66, 106)
125
(98, 159)
50
(40, 59)
0.60
(0.52, 0.63)
Zagreb 630,723 201
(138, 257)
348
(241, 444)
139
(104, 171)
0.69
(0.64, 0.77)
Amsterdam 616,727 41
(30, 68)
88
(66, 141)
20
(16, 26)
0.49
(0.30, 0.65)
Riga 550,511 45
(9, 95)
81
(15, 172)
31
(7, 57)
0.67
(0.54, 0.94)
Helsinki 453,721 27
(15, 39)
74
(42, 105)
22
(13, 31)
0.80
(0.75, 0.83)
Oslo 435,270 26
(7, 43)
88
(24, 142)
15
(5, 23)
0.55
(0.38, 0.71)
Vilnius 429,750 31
(5, 75)
79
(14, 197)
20
(2, 45)
0.68
(0.19, 0.93)
17 Grantham Institute Summer heat deaths in 854 European
cities more than tripled due to climate
change
Copenhagen 415,428 9
(5, 14)
25
(14, 40)
7
(4, 11)
0.82
(0.75, 0.87)
Bratislava 337,865 18
(9, 32)
68
(37, 123)
15
(8, 24)
0.85
(0.70, 0.93)
Tallinn 316,478 15
(5, 24)
52
(19, 86)
11
(4, 17)
0.77
(0.50, 0.9)
Ljubljana 217,685 35
(24, 46)
170
(116, 227)
26
(19, 34)
0.76
(0.70, 0.83)
Bern 176,221 54
(39, 66)
282
(203, 349)
33
(24, 41)
0.63
(0.51, 0.68)
Nicosia 171,474 16
(10, 21)
126
(81, 170)
10
(7, 12)
0.60
(0.56, 0.65)
Valletta 170,361 36
(25, 47)
256
(184, 337)
30
(22, 37)
0.83
(0.77, 0.89)
Luxembourg 70,074 7
(5, 9)
119
(88, 150)
4
(3, 5)
0.60
(0.48, 0.67)
2.3 Temporal variation
The top panel of Figure 7 shows the daily variation of the excess heat-related deaths under the
factual and counterfactual scenario across Europe over the summer. These results are based on an
aggregation of the 854 cities. The highest number of excess deaths is observed during the late
June - early July heatwave, whereas the second worst period was during the second week of
August. These two events correspond to the heat events labelled as ?box 1? and ?box 4? in Figure 4;
the daily estimates of the standardised excess deaths in cities of over 50,000 people in each
country are shown in Figure 8, with the same boxes overlaid for reference. As a result, the region
with the largest estimated numbers is western Europe, driven mainly by the extreme early heatwave
(Figure 7, lower left panel).
The effect of the Fennoscandia heatwave (Figure 8, box 2), although extremely impactful across the
affected countries as a whole (Barnes et al., 2025), led to only a moderate increase in health risks in
the large cities included in this study; however, as noted in section 1.1, the highest temperatures
were reported in the northern parts of the country, which are not represented in this sample, and
therefore the true risk is probably under-represented here.
The heatwave across the Balkans (Figure 8, box 3), which was associated with devastating wildfires
across the region, was associated with the highest daily risks of any of the events, with an average
of 10 additional deaths per million population across the seven-day period affected. Again, this is
almost certainly an underestimate of the true mortality during this event, because six Balkan
countries that fall outside of the scope of this study experienced very similar temperatures during
this period. Despite the relatively small spatial extent of this heatwave, the resulting increase in
mortality is clearly visible in the Europe-wide totals in Figure 7.
https://spiral.imperial.ac.uk/entities/publication/905c62f2-4268-45b7-8310-2de66a26a9e8
18 Grantham Institute Summer heat deaths in 854 European
cities more than tripled due to climate
change
Figure 7: Standardised excess deaths per million population, aggregated over all countries. Dark
blue boxes highlight periods of particularly dangerous heat that have previously been the subject of
attribution studies (see section 1.1 for details).
19 Grantham Institute Summer heat deaths in 854 European
cities more than tripled due to climate
change
Figure 8: Median estimates and 95% empirical confidence intervals of heat-related deaths under
factual and counterfactual temperature scenarios for the whole Europe and 4 major European
regions (eastern, southern, western and northern). The grey shaded area represents the estimated
number of heat-related deaths attributed to human-induced climate change. Note that the
estimates in this figure are based on aggregating the results of the 854 cities.
2.4 Comparison of results with previous literature
A study in Europe estimated that more than 60,000 people across Europe died from heat during
summer 2022 (Ballester et al., 2023). The same population and methods were used to estimate
heat-related deaths during the less intensely hot summer of 2023, reporting over 47,000 heat-
related deaths (Gallo et al., 2024). In our analysis, we report around 24,000 deaths across a
https://www.nature.com/articles/s41591-023-02419-z
https://www.nature.com/articles/s41591-024-03186-1?fromPaywallRec=false
20 Grantham Institute Summer heat deaths in 854 European
cities more than tripled due to climate
change
population of 158 million, approximately 30% of the 534 million population considered in that study,
placing our estimates in line with those of summer 2022. Our approach is also comparable with
another study focusing on the same 854 cities (Masselot et al., 2023), which reported an average of
13,589 deaths (11,530 to 15,475) between 1990 and 2019. Our estimate is about 10,000 heat-
related deaths higher, reflecting the much warmer summer of 2025 compared to the period
covered by that study.
Our study is also comparable with studies examining the proportion of heat-related deaths
attributed to human-induced climate change. A study on Europe focusing on summer 2022 found
that more than 50% of the excess deaths could be attributed to human-induced climate change
(Beck et al., 2024). A global study recently showed that more than 50% of the reported heatwave-
related deaths could be attributable to climate change (Hundessa et al., 2025). Our results suggest
that 68% of the observed heat-related deaths could be attributed to climate change. This
discrepancy may reflect differences in event definition, climate attribution approaches, as well as
the fact that our study focuses on urban populations, which generally experience greater warming
(Tuholske et al., 2021). Our results are comparable with our previous rapid attribution analysis in 12
European cities, which reported almost 65% of the observed excess deaths to be attributed to
human-induced climate change.
3 Data & methods
3.1 Data and methods used in temperature attribution
In section 1 of this report, we study the influence of anthropogenic climate change on average
summer temperatures by comparing the likelihood and intensity of similar weather conditions at
present with those in a 1.3°C cooler climate representing the preindustrial past (1850-1900, based
on the Global Warming Index). The data and methods used in this section are described in detail
below.
3.1.1 Observational datasets
The ?observed? temperatures for each city are extracted from the gridded datasets by selecting the
closest grid cell to the coordinates of the city.
The observational datasets used are as follows:
? ERA5 - The European Centre for Medium-Range Weather Forecasts's 5th generation
reanalysis product, ERA5, is a gridded dataset that combines historical observations into
global estimates using advanced modelling and data assimilation systems (Hersbach et al.,
2020). We use daily mean temperature data from the ERA5 product at a resolution of 0.25°
× 0.25°, from January 1950 until the end of August 2025.
https://pubmed.ncbi.nlm.nih.gov/36934727/
https://www.nature.com/articles/s41612-024-00783-2
https://www.sciencedirect.com/science/article/pii/S2666675825003133?via%3Dihub
https://www.pnas.org/doi/10.1073/pnas.2024792118
https://www.imperial.ac.uk/grantham/publications/all-publications/climate-change-tripled-heat-related-deaths-in-early-summer-european-heatwave.php
https://www.imperial.ac.uk/grantham/publications/all-publications/climate-change-tripled-heat-related-deaths-in-early-summer-european-heatwave.php
https://www.globalwarmingindex.org/
https://rmets.onlinelibrary.wiley.com/doi/full/10.1002/qj.3803
https://rmets.onlinelibrary.wiley.com/doi/full/10.1002/qj.3803
21 Grantham Institute Summer heat deaths in 854 European
cities more than tripled due to climate
change
? E-OBS (v31.0e + pre1950) - This is a gridded land-only dataset covering Europe, derived
from the interpolation of station?derived meteorological observations (Cornes et al., 2018).
We use daily mean temperature data from this product at a resolution of 0.25° × 0.25°, from
January 1920 to June 30th 2025. For three locations no data were available because the
nearest grid cell is not designated as land. For all other stations in the study, at least 62
complete summers of mean daily temperatures were used to construct the monthly time
series.
As a measure of anthropogenic climate change we use the global mean surface temperature
(GMST), where GMST is taken from the National Aeronautics and Space Administration (NASA)
Goddard Institute for Space Science (GISS) surface temperature analysis (GISTEMP, Hansen et al.,
2010 and Lenssen et al. 2019). To reduce variability in the GMST due to the El Nino-Southern
Oscillation (ENSO), we use a four-year rolling average of the GMST, centred on the third year.
3.1.2 Climate models
In the attribution step we use a multi-model ensemble of coupled global circulation models from
the CMIP6 experiments (Eyring et al., 2016). For all simulations, the period 1850 to 2014 is based on
historical simulations, while the SSP5-8.5 scenario is used to project from 2015 to 2030. All years
from 1850 to 2030 were used to estimate the historic warming trend. 38 models were used in the
analysis, each providing one realisation to the multi-model ensemble. In an exploratory analysis
comparing the spatial climatology of summer temperatures from 1990-2020 with the observed
climatology, all models were found to perform relatively well. To avoid using different subsets of
climate models in each region, no further selection was carried out based on the local seasonal
cycles or parameter fits. However, five model variants were removed from the original ensemble of
43, because they were lower-resolution variants of other models within the ensemble. Details of the
38 models included in the study can be found in Table A2 in the Appendix.
3.1.3 Statistical modelling of trends
Methods for observational and model analysis, and for synthesis of the results, are used according
to the World Weather Attribution Protocol, described in Philip et al., (2020), with supporting details
found in van Oldenborgh et al., (2021), Ciavarella et al., (2021), Otto et al., (2024) and here. The key
steps, presented in section 1, are: trend estimation from observations; attribution using climate
models; and synthesis of the attributable change in temperature.
The attribution analysis is carried out independently for each dataset, city and calendar month
(June, July and August). We first extract the time series of monthly mean temperatures by selecting
the grid cell closest to the coordinates of the city. For each time series we estimate the parameters
of a nonstationary normal distribution in which the mean monthly temperature depends linearly on
GMST, while the variance remains stationary. This statistical model is then used to estimate the
expected change in mean temperature associated with a 1.3°C reduction in GMST, corresponding
to the expected difference in temperature between the current climate and a 1.3°C cooler
https://doi.org/10.1029/2017JD028200
https://doi.org/10.1029/2010RG000345
https://doi.org/10.1029/2010RG000345
https://doi.org/10.1029/2018JD029522
https://gmd.copernicus.org/articles/9/1937/2016/
https://doi.org/10.5194/ascmo-6-177-2020
https://doi.org/10.1007/s10584-021-03071-7
https://doi.org/10.1007/s10584-021-03052-w
https://ascmo.copernicus.org/articles/10/159/2024/
https://www.worldweatherattribution.org/pathways-and-pitfalls-in-extreme-event-attribution/
22 Grantham Institute Summer heat deaths in 854 European
cities more than tripled due to climate
change
counterfactual climate representing the preindustrial past (1850-1900, based on the Global
Warming Index). 95% confidence intervals are obtained by bootstrapping, using a sample size of
1000.
3.1.4 Synthesis of trends
For a given city and calendar month, the estimated change in intensity from the observational
products and climate models are combined into a single result using WWA?s standard synthesis
algorithm (Otto et al., 2024). First, the results from observations are averaged, with a representation
error added in quadrature to reflect any difference in mean trend; the results from climate models
are combined into a separate precision-weighted average, with a term added in quadrature to
account for inter-model spread as well as internal variability. These two weighted averages are then
again combined using inverse-variance (precision) weighting, to obtain an overall best estimate and
estimated upper and lower bounds of a 95% confidence interval.
3.1.5 Calculating counterfactuals
Factual daily mean temperatures for June-August 2025 were taken from ERA5 reanalysis (E-Obs
data were not available for August at the time of writing). The preindustrial counterfactual was
obtained by subtracting the best estimate of the synthesised change in intensity for that day?s
calendar month from the factual temperature, and lower and upper confidence bounds by
subtracting the upper and lower bounds of the synthesised change, respectively. These three
counterfactual temperatures were used as inputs to the epidemiological model, along with the
factual temperatures, to estimate the change in mortality attributable to the synthesised change in
monthly mean temperatures.
3.2 Data & methods used in mortality attribution
To estimate heat-mortality attributable to human-induced climate change, we applied a framework
that combines exposure-response functions with temperature time series under factual and
counterfactual scenarios. The method comprises three key steps:
1. Deriving exposure-response functions: We obtained age-specific temperature?mortality
relationships for the 854 cities in Europe from Masselot, et. al., 2023 (freely available here).
2. Estimating heat-related mortality: These relationships were used to calculate excess
deaths under both factual and counterfactual temperature scenarios (see previous section).
3. Attributing deaths to climate change: The comparison (absolute or relative difference)
between excess deaths in the factual and counterfactual scenarios quantifies the impact of
human-induced climate change on mortality.
Step 1: Exposure-response functions were derived from daily mortality data from the Multi-country
Multi-city Collaborative Research Network (MCC; https://mccstudy.lshtm.ac.uk/), annual Eurostat
statistics, and additional city-specific data from sources including satellite observations (MODIS,
Copernicus). Daily mean temperatures came from the ERA5-Land reanalysis dataset for 1990?
https://www.globalwarmingindex.org/
https://www.globalwarmingindex.org/
https://ascmo.copernicus.org/articles/10/159/2024/
https://www.sciencedirect.com/science/article/pii/S2542519623000232?via%3Dihub
https://zenodo.org/records/10288665
https://mccstudy.lshtm.ac.uk/
23 Grantham Institute Summer heat deaths in 854 European
cities more than tripled due to climate
change
2019. The associations were stratified by age group (20?44, 45?64, 65?74, 75?84, 85+) and city,
capturing variations in vulnerability. This framework also applies pooling of the estimates, providing
robust curves in cities where data is sparse (Masselot, et. al., 2023). Figure 9 shows the relative risk
curves in the 30 European capitals available in the study.
Step 2: Relative risks were calculated for both factual and counterfactual temperatures and
converted into heat-related mortality fractions following Vicedo-Cabrera et al. (2023) and Beck et
al. (2024), accounting for lagged effects of heat. Daily deaths during the heatwave were estimated
assuming a constant rate based on historical Eurostat data (Masselot, et. al., 2023). Multiplying this
rate by the attributable mortality fraction yielded excess deaths in each scenario. As the interest
was in estimating the impact of heat, relative risks were derived from the temperature?mortality
curve focusing on temperatures above the minimum mortality temperature (MMT).
Step 3: Finally, the difference between excess deaths in the factual and counterfactual scenarios
provides the number of heat-related deaths attributable to anthropogenic climate change. We also
calculated the proportion of heat-related deaths in the factual scenario attributable specifically to
human-induced climate change (Beck et al., 2024).
Figure 9: Mean temperature-related relative mortality risk across the 30 European capitals by age
group, as retrieved from Masselot et al., 2023.
https://www.sciencedirect.com/science/article/pii/S2542519623000232?via%3Dihub
https://iopscience.iop.org/article/10.1088/1748-9326/ace0d0
https://www.nature.com/articles/s41612-024-00783-2
https://www.nature.com/articles/s41612-024-00783-2
https://www.sciencedirect.com/science/article/pii/S2542519623000232?via%3Dihub
https://www.nature.com/articles/s41612-024-00783-2
https://pubmed.ncbi.nlm.nih.gov/36934727/
24 Grantham Institute Summer heat deaths in 854 European
cities more than tripled due to climate
change
3.3 Limitations of the methods and sensitivity analyses
The rapid nature of this type of study precluded access to observed death counts across the 854
European cities. To address this, we assumed a constant daily death rate at the optimum
temperature in each city, based on historical annual mortality data. However, deaths typically peak
during heatwaves, so the true death rate is likely higher than the constant rate used here,
potentially leading to an underestimation of the absolute burden of heat. This limitation does not
affect relative estimates, as these are independent of the underlying mortality rate.
While this study examines all-cause mortality, it does not account for other adverse health effects
of heat, including impacts on mental health, hospital admissions, or medication use. The study also
does not account for adaptation. As the exposure-response functions used in this study were
derived using mortality data up to 2019, the results might not capture the potential temporal
attenuation of the temperature effect which has been reported in previous literature (Vicedo-
Cabrera et al., 2018). Indeed, improved adaptation policies and infrastructure (Vicedo-Cabrera et al.,
2018) and autonomous adaptation measures addressing the increasing burden of heat have the
potential to reduce vulnerabilities and impacts. In addition to this, our study does not capture the
changes in the baseline population that occurred post COVID-19. Not accounting for this might lead
to higher numbers in some cities.
We also note that temperatures in the cities may not be representative of the temperatures across
the whole country, so can?t extrapolate beyond the geographic scope of the study - this is
particularly the case in Scandinavia, where the cities are all in the south, while the hottest
temperatures were recorded further north. Furthermore, climate models are known to
underestimate the rate of warming across much of Europe (eg. Vautard et al., 2023) and so the
estimated changes in temperatures attributable to anthropogenic warming may themselves be
conservative estimates; the sensitivity of the analysis to this is considered in section 3.3.2 below.
3.3.1 Sensitivity to non-optimum temperatures
The main analysis focuses on temperatures above the MMT, excluding deaths from colder summer
temperatures. Table 4 extends the analysis to the full temperature-mortality curve. This increases
the total number of deaths due to non-optimal temperatures to 28,742 (26,311 to 31,089), as it also
includes cold-related mortality. In contrast, deaths attributable to human-induced climate change
are lower, 11,827 (10,091 to 13,438), nearly 5,000 fewer, but still substantial. The proportion
attributable to climate change remains high at 40% (36% to 46%). This reduction reflects a
protective effect in some regions, particularly northern Europe, where warming has lessened risks
from colder summer temperatures.
Table 4: Population figures, median and 95% empirical confidence intervals for excess mortality
related to non-optimal temperature presented as crude numbers and rates per 1 million population
for deaths attributable to heat, number of deaths attributable to human-induced climate change
and proportion of excess heat-related deaths that can be attributed to human-induced climate
https://pubmed.ncbi.nlm.nih.gov/29272855/
https://pubmed.ncbi.nlm.nih.gov/29272855/
https://pubmed.ncbi.nlm.nih.gov/29272855/
https://pubmed.ncbi.nlm.nih.gov/29272855/
https://www.nature.com/articles/s41467-023-42143-3
25 Grantham Institute Summer heat deaths in 854 European
cities more than tripled due to climate
change
change. Note that a minor rounding error occurs when summing the crude numbers by age group,
resulting in a slight difference from the overall total.
Age group Population Excess
deaths
Rate per 1
million
population
Attributable
to climate
change
Proportion
of excess
deaths due
to climate
change
Total 158,473,649 28742
(26311,
31089)
181
(166, 196)
11827
(10091,
13438)
0.41
(0.36, 0.46)
20-44 66,886,597 705
(572, 864)
11
(9, 13)
212
(119, 302)
0.30
(0.19, 0.39)
45-64 54,059,644 3547
(3111, 3978)
66
(58, 74)
1410 (1095,
1733)
0.40
(0.33, 0.46)
65-74 19,676,737 4426
(3979, 4829)
225
(202, 245)
1785
(1465, 2085)
0.40
(0.35, 0.46)
75-84 12,762,858 8264
(7555, 8985)
647
(592, 704)
3486
(2973, 3941)
0.42
(0.37, 0.47)
85+ 5,087,812 11785
(10824,
12830)
2316
(2127, 2522)
4960
(4159, 5687)
0.42
(0.36, 0.47)
3.3.2 Sensitivity to strength of temperature trend
It is likely that the synthesised changes in temperature presented in section 1.2 represent a
conservative estimate of the true level of warming in many regions across Europe.. Figure 10 shows
a map of the historical change in temperatures in the 854 cities associated with 1.3°C of warming,
obtained by averaging the change estimated from two observational datasets: ERA5 (reanalysis
data based on a forecasting model, from 1950-2025) and E-Obs (interpolated from observations at
weather stations, 1920-2024). Figure 6b shows the average warming simulated by 38 CMIP6
climate models, and figure 6c the difference between the two, highlighting where the
underestimation is greatest. While the spatial pattern of warming across the continent is fairly
similar in models and observations, the climate models fail to simulate the full range of observed
changes: in observations, the monthly changes range from 0.16 to 5.1°C, and in the climate models,
from 1.1 to 2.5°C (Figure 6d, with the distribution of observed changes in blue and modelled
changes in red). This is particularly the case in June, which has the widest spread of observed
changes and the narrowest spread of modelled changes, but is true of all months. The result of this
is that for June temperatures, the synthesised changes (pink) are on average 0.8°C cooler than
those actually observed (95% central interval of differences: -1.9°C to +0.3°C); for July, 0.4°C cooler
(-1.3°C to +0.2°C); and for August, 0.5°C cooler (-1.3°C to +0.2°C). While some of this difference
may be due to natural variability in the real world, previous studies have highlighted similar
26 Grantham Institute Summer heat deaths in 854 European
cities more than tripled due to climate
change
discrepancies in climate models over Europe (eg. Vautard et al., 2023). The estimated temperature
changes here should therefore be interpreted as a conservative estimate of the true level of human-
caused warming across Europe.
To test the sensitivity of the mortality attribution to the choice of dataset, the analysis was run using
not only the best estimate but also the upper and lower bounds of a 95% confidence interval for the
change in each city and month, and repeated using trends estimated from observations only. The
results of this analysis are shown in Table 5. Use of the observed temperature increases resulted in
a larger number of heat-related deaths attributable to climate change, (18,825; 17,115 to 20,437),
and a larger proportion (77%; 74% to 80%). Comparing these results to those shown in Table 1, we
see that the lower bound of the number of attributable deaths when using observations only to
estimate the trend overlap with the upper bound of the number of deaths when using the
synthesised results.
https://www.nature.com/articles/s41467-023-42143-3
27 Grantham Institute Summer heat deaths in 854 European
cities more than tripled due to climate
change
Figure 10: Maps of estimated change in summer temperatures in (a) observational data products
and (b) CMIP6 climate models; (c) differences between the two data sources. (d) Boxplot and
violinplots of the estimated changes each month across all cities, in observations (blue) and models
(red), with synthesised change in pink. The box represents the central 50% of the distribution, the
whiskers indicate points within 1.5 times the interquartile range of the box, and circles indicate
points more extreme than this.
Table 5: As Table 1, but using only observed changes in temperature to create the counterfactual
daily temperatures, rather than the combined output of observations and climate models.
Population figures, median and 95% empirical confidence intervals for excess mortality related to
heat (temperatures higher than the minimum mortality temperature) presented as crude numbers
and rates per 1 million population for deaths attributable to heat, number of deaths attributable to
human-induced climate change and proportion of excess heat-related deaths that can be attributed
to human-induced climate change. Note that a minor rounding error occurs when summing the
crude numbers by age group, resulting in a slight difference from the overall total.
Age group Population Excess
deaths
Rate per 1
million
population
Attributable
to climate
change
Proportion
of excess
deaths due
to climate
change
Total 158,473,649 24404
(21968,
26806)
154
(139, 169)
18825
(17115, 20437)
0.77
(0.74, 0.80)
20-44 66,886,597 608
(434, 795)
9
(6, 12)
328
(245, 415)
0.54
(0.44, 0.66)
45-64 54,059,644 3058
(2557, 3593)
57
(47, 66)
2150
(1833, 2458)
0.71
(0.64, 0.77)
65-74 19,676,737 3738
(3279, 4171)
190
(167, 212)
2839
(2528, 3122)
0.76
(0.72, 0.80)
75-84 12,762,858 7019
(6305, 7726)
550
(494, 605)
5505
(5011, 5982)
0.78
(0.75, 0.81)
85+ 5,087,812 9959
(9012, 10964)
1957
(1771, 2155)
7993
(7276, 8717)
0.80
(0.77, 0.83)
28 Grantham Institute Summer heat deaths in 854 European
cities more than tripled due to climate
change
Appendices
Figure A1: Synthesised change in temperatures per calendar month
Table A1: Country specific population figures (based on the available cities per country), median
and 95% empirical confidence intervals for excess heat-related presented as crude numbers, age-
standardised rates (using the European standard population) per 1 million population, number of
deaths attributable to human-induced climate change and proportion of excess heat-related deaths
that can be attributed to human-induced climate change. The numbers reflect the heat-related
impacts during the entire summer of 2025.
Country Population Excess
deaths
Standardise
d excess
mortality per
million
Attributable
to climate
change
Proportion
of excess
deaths due
to climate
change
United
Kingdom
26,545,907 1687
(1303, 2045)
72
(56, 87)
1147
(900, 1384)
0.68
(0.67, 0.69)
Germany 23,406,685 2445
(2025, 2918)
101
(83, 121)
1477
(1245, 1701)
0.60
(0.54, 0.65)
France 17,962,199 2062
(1710, 2406)
110
(91, 129)
1444
(1224, 1661)
0.71
(0.66, 0.73)
Spain 17,714,878 3893
(3264, 4753)
207
(173, 253)
2841 (2374,
3310)
0.72
(0.65, 0.78)
Italy 17,186,805 6710
(5828, 7649)
335
(290, 382)
4597
(4083, 5137)
0.68
(0.64, 0.73)
Poland 10,396,899 694
(422, 1092)
75
(46, 117)
543
(353, 784)
0.78
(0.70, 0.85)
Romania 6,010,799 1487
(1031, 1954)
283
(198, 372)
1064
(767, 1342)
0.72
(0.66, 0.79)
Netherlands 5,644,325 411
(288, 665)
79
(56, 126)
200
(141, 257)
0.49
(0.29, 0.65)
29 Grantham Institute Summer heat deaths in 854 European
cities more than tripled due to climate
change
Portugal 3,457,781 386
(314, 494)
108
(88, 138)
217
(180, 261)
0.56
(0.5, 0.61)
Greece 3,401,064 1372
(1127, 1597)
380
(312, 442)
808
(697, 913)
0.60
(0.54, 0.64)
Sweden 2,807,484 82
(49, 116)
30
(18, 43)
67
(41, 95)
0.83
(0.76, 0.87)
Hungary 2,806,295 362
(225, 552)
139
(87, 212)
266
(178, 366)
0.73
(0.65, 0.83)
Bulgaria 2,596,877 866
(620, 1115)
371
(267, 477)
552
(404, 683)
0.64
(0.57, 0.71)
Czechia 2,522,881 144
(88, 224)
66
(41, 102)
113
(71, 163)
0.79
(0.7, 0.84)
Belgium 2,455,390 269
(216, 324)
111
(89, 134)
154
(124, 182)
0.58
(0.50, 0.62)
Switzerland 1,977,029 280
(208, 355)
145
(107, 184)
207
(154, 252)
0.75
(0.59, 0.81)
Austria 1,949,160 180
(134, 239)
102
(75, 135)
136
(103, 170)
0.76
(0.67, 0.83)
Finland 1,460,602 105
(43, 166)
76
(32, 121)
67
(24, 110)
0.65
(0.47, 0.74)
Croatia 1,101,531 394
(296, 483)
382
(289, 469)
268
(210, 319)
0.68
(0.63, 0.74)
Ireland 1,090,316 13
(5, 22)
18
(7, 30)
9
(4, 15)
0.72
(0.70, 0.75)
Lithuania 1,033,748 74
(15, 173)
73
(15, 170)
50
(10, 109)
0.70
(0.37, 0.90)
Denmark 917,663 22
(13, 31)
27
(16, 38)
17
(10, 23)
0.77
(0.72, 0.80)
Slovakia 877,322 62
(36, 103)
94
(56, 155)
51
(32, 77)
0.83
(0.73, 0.91)
Latvia 874,434 73
(16, 156)
83
(18, 178)
49
(13, 94)
0.67
(0.56, 0.92)
Norway 829,551 52
(23, 77)
81
(36, 119)
30
(14, 43)
0.58
(0.48, 0.66)
Cyprus 457,113 49
(36, 61)
146
(106, 182)
29
(22, 35)
0.59
(0.55, 0.63)
Estonia 440,155 24
(9, 41)
58
(22, 98)
16
(6, 27)
0.70
(0.44, 0.86)
Slovenia 308,320 65
(47, 83)
223
(162, 287)
43
(33, 54)
0.66
(0.61, 0.74)
Malta 170,361 36
(25, 47)
256
(184, 337)
30
(22, 37)
0.83
(0.77, 0.89)
Luxembourg 70,074 7
(5, 9)
119
(88, 150)
4
(3, 5)
0.60
(0.48, 0.67)
30 Grantham Institute Summer heat deaths in 854 European
cities more than tripled due to climate
change
Authors
Clair Barnes, Centre for Environmental Policy, Imperial College, London, UK
Garyfallos Konstantinoudis, Grantham Institute for Climate Change and the Environment, Imperial
College, London, UK
Pierre Masselot, Environment & Health Modelling Lab, London School of Hygiene and Tropical
Medicine, London, UK
Malcolm Mistry, Environment & Health Modelling Lab, London School of Hygiene and Tropical
Medicine, London, UK
Antonio Gasparrini, Environment & Health Modelling Lab, London School of Hygiene and Tropical
Medicine, London, UK
Ana M Vicedo-Cabrera, Institute of Social and Preventive Medicine, Oeschger Center for Climate
Change Research, University of Bern, Bern, Switzerland
Emily Theokritoff, Grantham Institute for Climate Change and the Environment, Imperial College,
London, UK
Ben Clarke, Centre for Environmental Policy, Imperial College, London, UK
Friederike Otto, Centre for Environmental Policy, Imperial College, London, UK
Review authors
Sjoukje Philip, Royal Netherlands Meteorological Institute (KNMI), De Bilt, The Netherlands
Mariam Zachariah, Centre for Environmental Policy, Imperial College, London, UK
Media enquiries: grantham.media@imperial.ac.uk
mailto:grantham.media@imperial.ac.uk
31 Grantham Institute Summer heat deaths in 854 European
cities more than tripled due to climate
change
Please cite as:
Barnes, C., et al. (2025). Summer heat deaths in 854 European cities more than tripled due to
climate change. Grantham Institute report.
DOI: https://doi.org/10.25560/123873
This work is licensed under a Creative Commons Attribution-Non Commercial-No Derivatives 4.0
International License.
https://doi.org/10.25560/123873
