The objective of this study is to investigate the climate sensitivity of health in a moder- ate climate of SW Germany. Daily mortality rates for the 30 yr period 1968-1997 for Baden-Württem- berg (SW Germany) have been investigated with regard to the possible impacts of the thermal envi- ronment. A complete heat budget model of the human being (Klima-Michel model with outcome 'perceived temperature') has been used to assess the atmospheric conditions of heat exchange. Mor- tality data show a marked seasonal pattern with a minimum in summer and a maximum in winter. During the seasonal minimum in summer, death rates rise sharply with increasing heat load, reach- ing highest values during pronounced heat waves. Under comfortable conditions, mortality data show the lowest rates. Increasing cold stress also causes death rates to rise. In addition, thermal changes on a time scale of 1 wk have been considered in comparison to short-term exposures. In all seasons changes towards 'warmer' conditions in terms of perceived temperature result in adverse effects, while changes to 'colder' conditions provide relief. This is unexpected for the winter. The daily correlation coefficients between the deviations of perceived temperature and the deviations of mortality rate from the smoothed values (Gaussian filter, 101 d) show a pronounced seasonal pattern with significant differences from zero between March and August. From the end of June to the begin- ning of July, about 25% of the variance in the deviations of mortality rate from the smoothed values can be explained by the effects of the thermal environment. The winter values show only non-signif- icant correlations, strong day-to-day variability, but marked time lags of 8 d and more, while in sum- mer there is practically no difference in the results between the zero and 1 d lags. Cold spells lead to excess mortality to a relatively small degree, which lasts for weeks. The mortality increase during heat waves is more pronounced, but is followed by lower than average values in subsequent weeks.