BACKGROUND: Ambient PM(10) is associated with mortality; however, potential changes in this association over time and the factors explaining such changes are unclear. Therefore, we aimed to examine whether mortality risk associated with PM(10) has changed from 1979 to 2019 and whether changes in socioeconomic or environmental conditions can explain any temporal variation in the association between PM(10) and mortality. METHODS: We applied an extended two-stage time-series design to assess temporal change in the association between PM(10) and all-cause mortality across 143 cities in 26 countries from 1979 to 2019. In the first stage, city-specific and time-specific associations between PM(10) and mortality were estimated using quasi-Poisson regression after each city time series was divided into non-overlapping 3-year segments. In the second stage, these estimates were pooled by use of longitudinal random-effects meta-regression with calendar year as a predictor. We further investigated whether selected socioeconomic and environmental factors explained observed temporal trends by including these variables in the second-stage model. FINDINGS: Totally, 23·2 million deaths were analysed. The overall association between PM(10) and mortality had increased from 1979 to 2019, indicating a stronger association at a given PM(10) concentration over time. A 10 μg/m(3) increase in daily PM(10) was associated with a 0·23% increase in all-cause mortality in 1979 (95% CI 0·05-0·41), and this association increased to 0·51% in 2019 (0·36-0·65). Temporal patterns in the PM(10)-mortality association varied across cities and were positively associated with population ageing over time and negatively associated with annual mean PM(10) concentrations. INTERPRETATION: The findings of this study suggest that the effect of a given increment of PM(10) on mortality has increased over time. Applying historical risk estimates could underestimate the current health burden. Continuous updating of evidence on the health impacts of air pollution is essential to ensure accurate and valid estimates. FUNDING: Wellcome Trust.