This study was conducted in order to explore the effect of thermal comfort on all-cause mortality using three indices in different lag times, in a semi-arid to dry region of Iran. Three thermal comfort indices based on the energy balance of the human body including physiologically equivalent temperature (PET), predicted mean vote (PMV), and standard effective temperature (SET) were used to assess the effects of thermal comfort on mortality. Distributed lag non-linear models were used to assess the relation. The natural cubic spline was chosen as the basis function for the space of predictors and lags, with 4 degrees of freedom. All three indices showed the same pattern in general, but the relative risk for PMV values were more than the other indices in different lags. For all three indices, lag 0 had the highest relative risk of mortality in warm and hot indices. The relative risk for warm and hot values was more than cool and cold values in lag 0, and for the PMV index, it was larger than the two other indices. These results were different in lags 5 to 8, and the relative risks for cool and cold values were more than warm and hot values. This study showed that heat stress has a stronger and more immediate adverse effect on mortality than cold stress. Also, the elderly and females are more vulnerable than others. The most apparent effect was seen in lags 0-12.