Background: Research has failed to find the expected clear-cut differe
nce in the presence of events provoking onset in endogenous and nonend
ogenous depression. Methods: A longitudinal study of 127 depressed fem
ale patients from two psychiatric departments were studied using the P
resent State Examination and the Life Event and Difficulty Schedule. T
wo earlier patient series using the same measures were employed to che
ck findings, and two general population series were used to estimate t
he expected rate of life events. Results: A large proportion of patien
ts experienced a severely threatening event before onset, with the exc
eption of a group defined by a British diagnostic category (a relative
ly small group of patients with melancholic/psychotic depression who w
ere not experiencing their first episode). The results were broadly re
plicated in the two other patient series. The proportions of patients
who experienced ongoing major difficulties did not differ between the
groups. Conclusions: The relative size of this melancholic/psychotic g
roup of patients with a prior onset may well have varied markedly from
study to study in previous research, and this may help to explain the
puzzling variability in findings concerning the role of stressful eve
nts in endogenous depression.