Longitudinal research in Stirling County, Atlantic Canada, indicated that during the 1950s and 1960s the prevalence of depression was significantly and persistently higher in the "low" socioeconomic status population than at other socioeconomic status levels. Anxiety was found to show a less clear picture. Incidence of depression after the study started was also higher among those who were initially in the low socioeconomic status group, supporting the view that the stress of poverty may be causally related to depression. There was also a trend for prior depression to be associated with subsequent downward social mobility, supporting the view that the concentration of depressed people at the lower end of the social hierarchy may result from handicapping aspects of the illness. Neither of these trends was statistically significant. More striking was evidence that, irrespective of socioeconomic status, depression carried a substantial risk for poor clinical course and outcome. Both depression and poverty tended to be chronic, and, accordingly, their association at the end of the study was influenced by their association at its beginning. The stability of the relationship between poverty and depression warrants the attention of caregivers and policymakers and raises new questions about strategies for the study of causal sequences.