A statistically significant season of birth variation is found in an unselected nationwide sample of 1,939 eating disorders patients, with peak season of birth occurring in May. However, among younger patients (n = 882), peak season of birth is in March, which is statistically significantly different to that expected from the general population season of birth cycle. This finding may imply links between etiology of earlier-onset eating disorders and the psychoses; similar first quarter peak seasonal patterns of birth have been found in schizophrenic and affective psychoses--with birth peaks in January and February. In contrast, for the neuroses and personality disorders, birth peaks have been found to be in June and August, similar to the June birth peak found in this study for later-onset eating disorders (n = 1,057), which was not statistically significantly different to season of birth peaks expected from general population data.