Few studies have examined whether risk factors for depressive symptomatology differ in bereaved and nonbereaved individuals or whether risk factors differ in bereaved individuals over time. Between 1979 and 1983 in Washington County, Maryland, the associations between various health and social network variables and depressive symptomatology were evaluated prior to bereavement and at 1 and 12 months after bereavement in 136 widows and 409 married controls. Prior Center for Epidemiologic Studies Depression Scale scores were generally a good predictor of subsequent scores; however, shortly after bereavement prior scores proved relatively uninformative as most widows experienced a marked increase in depressive symptomatology. Poor health and limitations in physical activity at baseline were consistently associated with higher levels of symptomatology. Although having more friends was also consistently associated with lower levels of symptomatology, the effect of family size appeared to be time and circumstance specific. These results suggest that women at risk of prolonged depression after the death of their husbands can be identified prior to or at the time of bereavement and that widows have risk factors similar to those of women at risk of depression in the general community.