A special analysis of mortality data for England and Wales fof 1972 has been used to examine the relationship between month of birth and month of death. For persons 75 and over in eight subgroups (by sex and marital status) there was a consistent trend in deaths with an excess in the birth month and the following three months. This excess was of the order of only 1% of all deaths in the year but statistically significant for each subgroup. The method of analysis was adjusted for the nonsynchronous monthly variation in births and deaths, but a systematic error in recording the month of birth could not be excluded. A wide range of studies on the relationship between stress, morbidity, and mortality are briefly reviewed; it is suggested that in the elderly a birthday may in some subtle way influence the general morale of an individual. Further studies have been planned to test whether 'birthday stress' is a realistic explanation, or if a number of other hypotheses need to be invoked.