A summary of the literature shows that there are well-marked seasonal variations in mortality for total deaths, for respiratory and cardiovascular diseases and for diabetes, the mortality rate being higher in winter than in summer. These seasonal variations in mortality are seen in infants under 1 year of age and in older people but not in youths and young adults. The amplitude of the seasonal variation in mortality increases with increasing age because of the higher incidence of cardiovascular and respiratory mortality in older people. Seasonal variations in air temperature are a more important determinant of seasonal variations in respiratory and cardiovascular deaths than are fluctuations in air pollution; there is usually a time-lag of up to a week in the change in air temperature before the mortality rate for these diseases increases; a run of 4 - 5 days of stressful temperatures (either hot or cold) has more effect on mortality than an isolated hot or cold day. Examination of the seasonal variations in all causes of death of the four population groups in South Africa shows that Whites and Asians display the typical pattern of a winter high and a summer low mortality of populations in developed countries. The seasonal variation in mortality of Coloureds and Blacks is quite different. It shows two peaks, one in winter and one in summer. This bimodal pattern in the seasonal variation is due to the fact that one-third of all Coloured and Black deaths occur in infants under 1 year of age and most of these deaths occur in summer as a result of gastro-enteritis. Comparison was made of the seasonal variations in mortality rates for all causes of death and for respiratory and ischaemic heart disease (IHD) deaths of Whites over 40 years of age in Durban and Johannesburg. This showed that the seasonal variation for all causes of death is greater in Johannesburg than in Durban, i.e. proportionately more older Whites die during winter in Johannesburg than in Durban. The reason for this difference is that the seasonal variation in respiratory disease mortality is much greater in Johannesburg than it is in Durban, but the seasonal variation in IHD mortality is the same in the two cities.