Background. The stress generation hypothesis proposed by Hammen (1991) holds that depressed individuals generate stressful conditions for themselves, which lead to recurrence. The original test of this hypothesis compared dependent life events in women with recurrent depression to medical and normal controls. Two further research questions emerged from this work: (a) do individuals with a history of many depressive episodes generate more dependent life events than depressives with fewer episodes?; and (b) what is the aetiological relevance of any stress that may be generated?

Methods. The present research tested differences in dependent and independent events between depressed individuals who had experienced: (a) no previous major depressive episodes; (b) one previous episode; and (c) two or more previous episodes. We predicted that, based on the stress generation hypothesis, recurrent depressives would show more dependent events than people without a depression history, and that these generated stressors would be of aetiological importance for precipitating recurrence (i.e. severe events in the 3 months preceding recurrence).

Results. Recurrent depressives experienced significantly more total dependent events than first-onset depressives in the 12 months, but not the 3 months, preceding their episode.

Conclusions. Although the findings supported the general premise of stress generation, the aetiological relevance of the generated stress for recurrence requires further study.