The aim of the study was to establish to what extent extreme thermal conditions have changed and how they affected mortality, and what conditions favor lower mortality rates or conversely, higher mortality rates. Heat/cold exposure was measured with the Universal Thermal Climate Index (UTCI). Daily mortality and meteorological data for 8 large Polish cities (Białystok, Gdańsk, Kraków, Lublin, Łódź, Poznań, Warszawa, and Wrocław) in the period 1975-2014 were analyzed. Generalized additive models were used to investigate the relationship between UTCI and mortality, and TBATS models were used for the evaluation of time series UTCI data. Most of the cities experienced a clear and statistically significant at p ≤ 0.05 decrease in cold stress days of 0.8-3.3 days/year and an increase in the frequency of thermal heat stress days of 0.3-0.6 days/year until 1992-1994. There was a clear difference as regards the dependence of mortality on UTCI between cities located in the "cooler" eastern part of Poland and the "warmer" central and western parts. "Cool" cities were characterized by a clear thermal optimum, approx. in the range of 5-30 °C UTCI, changing slightly depending on cause of death, age, or sex. For UTCI over 32 °C, in most of the cities except Gdańsk and Lublin, the relative risk of death (RR) rose by 10 to 20%; for UTCI over 38 °C, RR rose to 25-30% in central Poland. An increase in mortality on cold stress days was noted mainly in the "cool" cities: RR of total mortality increased even by 9-19% under extreme cold stress.