In recent years, theorists and researchers have disagreed about the relationship between social support and mental health. Some believe that support is a direct provoking agent (i.e. lack of support constitutes strain), whereas others maintain that support is a vulnerability factor moderating the effect of life stress. Focusing on clinical depression, the article reviews the arguments and evidence supporting a strain hypothesis of social support versus a vulnerability hypothesis. Reanalyzing cross-classified data from 12 community studies of clinical depression, the study shows that the choice of model depends on the specification of functional form of the stress-clinical depression relationship. The linear probability specification suggests a vulnerability hypothesis, whereas the logit and probit specifications support a strain hypothesis. However, theoretical and statistical arguments tend to favor a logit or probit specification, and an additional analysis of data from Brown and Harris [Social Origins of Depression: A Study of Psychiatric Disorder in Women. The Free Press, New York, 1978] supports these arguments. Thus, the study concludes that the strain hypothesis of social support is more consistent with the available data.