The summer of 2026 brought a relentless series of extreme heatwaves to Western Europe, breaking monthly and annual temperature records repeatedly across many countries from May onwards. Behind the record charts and emergency alerts lies a stark numerical accounting: an estimated 50,391 heat-related excess deaths occurred across 854 European cities between June and August, and around two-thirds of those deaths are attributable to anthropogenic climate change.
The study, led by Clair Barnes and Pierre Masselot with colleagues including Friederike Otto, Antonio Gasparrini and Ana Maria Vicedo-Cabrera, combined probabilistic climate attribution with city- and age-specific temperature–mortality relationships. The severity of the summer was such that several government agencies took the unprecedented step of issuing near-real-time estimates of the mortality associated with the most intense heat events, reflecting what the authors describe as the severity of the threat to public health. The new analysis provides the systematic, continent-wide counterpart to those emergency estimates, comparing the summer as it actually unfolded against a counterfactual climate representing a world without human-caused warming.
The methodological framework proceeds in two stages. In the first, the researchers estimated the local warming in each city attributable to human activity, then subtracted that warming from observed, factual temperatures to simulate counterfactual temperatures representative of a cooler, pre-industrial-influenced climate. Factual temperatures were drawn from the ERA5-Land reanalysis, with counterfactuals obtained by subtracting the estimated attributable local warming from those observed values. The degree of local warming was estimated using an established probabilistic attribution methodology: daily mean temperature series for each of the 854 cities during the Northern Hemisphere summer were retrieved both from observations-based datasets — ERA5-Land reanalysis from 1950 to 2026 and E-Obs interpolated observations from 1920 to 2026 — and from 36 distinct models in the CMIP6 multimodel ensemble, using historical forcings from 1850 to 2014 and SSP5-8.5 forcings from 2015 to 2030.
A regression model estimated the local warming in each city in response to each degree of global warming, allowing the team to calculate the amount of local warming associated with the 1.4 degrees Celsius of human-caused global warming estimated to have occurred by 2026. These city-level estimates were combined into an observational synthesis and a climate model synthesis, and the researchers took the unweighted mean of the two to avoid over-weighting towards climate models, which are known to underestimate local warming in Europe.
The second stage of the framework was a formal health impact assessment. Heat impacts were estimated using cumulative exposure–response functions derived from distributed lag nonlinear models, trained on historical mortality data from 2000 to 2019 in 232 EU cities. Crucially, the resulting functions were age specific — covering years 20–44, 45–64, 65–74, 75–84 and 85+ — accounting for the unequal burden that extreme heat places on older populations. The functions were combined with the factual and counterfactual temperature time series to estimate the relative risk of heat mortality under both scenarios.
To convert relative risks into actual death counts, the team used historical mortality data from Eurostat and population data from WorldPop to estimate mortality rates for 2026 across the 854 cities, integrating these rates with the modelled relative risks to estimate the number and proportion of heat-related excess deaths attributable to human-induced climate change. Uncertainty was propagated rigorously: the researchers constructed 1,000 spatially coherent samples of counterfactual temperatures by sampling from a multivariate normal distribution matching the synthesized attributable warming in each city and calendar month, and assessed uncertainty in the exposure–response functions through Monte Carlo simulation with 1,000 samples from the parameters of the fitted meta-regression model. Medians and 95% empirical confidence intervals for all quantities were computed from these 1,000 samples, capturing uncertainty about both the rate of local warming and the mortality functions.
The headline result is sobering. Of the estimated 50,391 heat-related excess deaths across the 854 cities during summer 2026, around two-thirds were attributable to anthropogenic climate change. In other words, in a counterfactual world without the human-caused warming that has accumulated since the pre-industrial era, roughly two out of every three of those deaths would not have occurred. The authors emphasize that these results highlight the threat from human-caused warming and the need for rapid emissions reductions alongside preparation and adaptation measures — a dual message that the record-breaking summer made impossible to ignore.
To test how sensitive their estimates were to the choice of population and mortality rates, the researchers repeated the analysis using the mean population and annual mortality rates from 2000 to 2019, the same assumptions used in a companion study covering previous years. As expected, the absolute estimates of excess deaths were generally slightly lower because of historically smaller population sizes, but the proportion of heat-related deaths attributable to anthropogenic warming remained stable.
The sensitivity analysis also enabled a direct historical comparison that puts 2026 into chilling context. Under the same assumptions about population size and annual mortality rates, the earlier work found that the summers with the highest heat-related mortality were 2022, with 29,480 deaths under the factual scenario (95% empirical confidence interval 26,643–32,006), and 2003, with 28,462 deaths (95% empirical confidence interval 25,709–31,199). Applying identical assumptions to summer 2026, the researchers estimate 45,492 deaths (95% empirical confidence interval 41,429–48,719) — around 50% higher than in either 2022 or 2003. The modelled number of heat-related deaths attributable to human-caused warming in 2026 alone, 29,330 (95% empirical confidence interval 23,495–34,420), is of similar magnitude to the total number of deaths due to extreme heat in either of those two landmark deadly summers.
The 2026 estimate covers 854 cities, and the authors note that the approach has strengths and limitations discussed in detail in the referenced methodological work; the exposure–response functions were trained on EU cities, and the attribution rests on modelled counterfactual temperatures rather than observed ones.
Subject of Research: Attribution of heat-related urban mortality in Europe during summer 2026 to human-induced climate change
Article Title: Heat-related urban mortality attributable to human-induced climate change in Europe in summer 2026
Article References: Barnes, C., Masselot, P., Xi, D., Rayner, A., Chen, X., Raju, E., Mistry, M., Gasparrini, A., Otto, F., Vicedo-Cabrera, A. M., & Konstantinoudis, G. (2026). Heat-related urban mortality attributable to human-induced climate change in Europe in summer 2026. Nature Health. https://doi.org/10.1038/s44360-026-00220-z
Image Credits: AI Generated
DOI: 10.1038/s44360-026-00220-z
Keywords: heat mortality, climate attribution, Europe, anthropogenic climate change, heatwaves, urban health, exposure-response functions, counterfactual climate, CMIP6, excess deaths, Nature Health, adaptation
News Source: Ophelia Keating. (November 4, 2026). Human-caused warming drove two-thirds of Europe’s 50,391 summer heat deaths in 854 cities. Scienmag.
