Three climatically diverse regions were studied to determine the impact of temperature change on heat-related mortality from 1971 to 1997. Median regressions showed that May-August temperatures in North Carolina rose by 1.0 degrees C (95% CL 0.0-2.0 degrees C) from 23.5 degrees C (74.3 degrees F), were unchanged in South Finland at 13.5 degrees C (56.3 degrees F), and rose in Southeast England 2.1 degrees C (0.3-4.0 degrees C) from 14.9 degrees C (58.8 degrees F). After determining for each region the daily temperature (as a 3 degrees C band) at which the mortality was the lowest, annual heat-related mortality was obtained as excess mortality per million at temperatures above this. Annual heat-related mortality per million (among the population at risk, aged 55+) fell in North Carolina by 212 (59-365) from 228 (140-317) to only 16 (not significant, NS); fell in South Finland by 282 (66-500) from 382 (257-507) to 99 (NS); and fell in Southeast England by 2.4 (NS) from 111 (41-180) to 108 (41-176). The falls in North Carolina and South Finland remained significant after allowances were made for changes in age, sex, and baseline mortality. Increased air conditioning probably explains the virtual disappearance of heat-related mortality in the hottest region, North Carolina, despite warmer summers. Other lifestyle changes associated with increasing prosperity probably explain the favorable trends in the cooler regions.