Although stressful life events (SLEs) play a major role in many etiologic theories of major depression (MD), important questions remain about the nature of their association with the onset of depressive episodes. We assessed over the last year, in female twins ascertained from a population based registry, the occurrence of 15 classes of SLEs and the onset of DSM-III-R MD and 2-week generalized anxiety disorder (GAD). The sample contained 24,648 person-months, 316 onsets of MD, and 239 onsets of GAD. SLEs were rated on long-term contextual threat and dependence. Discrete time-survival analyses were employed. The association between SLEs and depressive onsets was usually strongest in the month of occurrence but extended for "difficulty-like" events for up to 6 months. The depressogenic effect of SLEs was strongly predicted by contextual threat level, although some low threat events significantly increased risk for MD. The risk for a depressive onset given the number of reported SLEs within one month was: no event, 0.9%; one, 3.4%; two, 6.8%; and three, 23.8%. Although a few events were relatively specifically depressogenic or anxiogenic, most SLEs increased risk for both MD and GAD. The risk period produced by SLEs range from short-lived to relatively prolonged. High threat events encompass most but not all of the depressogenic effects of SLEs. Multiple SLEs in the same month substantially increase the risk for a depressive onset. The specificity of most SLEs for depressive versus anxiety syndromes is modest.