Poland is in the group of European countries with the most severe air quality and specific emission structure (over 80 % of PM2.5 emitted in residential combustion). In this work, we quantify the health impacts of PM2.5 concentrations in Poland using two chemical transport models (CTMs), WRF-Chem and EMEP4PL. The models were run for two years (2017 and 2018), which differed in meteorological conditions. The results show that the total number of premature deaths that could be avoided if the PM2.5 concentrations do not exceed the WHO recommended value of 5 μg m−3 exceeds 22,000 deaths for all model runs and meteorological years considered, with the lowest estimate at 22,447 (WRF-Chem model run for the year 2017; 95 % confidence intervals CI (95 % CI) 17121–25045) and the highest at 29,227 (95 % CI 22343 – 32572; EMEP4PL for the year 2018). These numbers are underestimated, as both CTMs underestimate the measured PM2.5 concentrations at site locations. The underestimation is the highest during the heating seasons and can be reduced if the emission of primary PM2.5 from residential combustion increases. The total number of estimated premature deaths is sensitive to chemical transport model selection and meteorological year. The selection of the meteorological year for the analysis may change the estimated number of premature deaths by 10 %. This change is significant compared to the estimated change in health impact due to the envisaged implementation of the Clean Air Program in Poland.