The associations between social relationships and health have been examined using two major research traditions. Using a social epidemiological approach, much research has shown the beneficial effect of social supports on health and health behaviors. Family interaction research, which has grown out of a more clinical tradition, has shown the complex effects of family functioning on health, particularly mental health. No studies have examined the relative power of these two approaches in explicating the connections between social relationships and health. We hypothesized that social relationships (social support and family functioning) would exert direct and indirect (through depressive symptoms) effects on health behaviors. We also hypothesized that the effects of social relationships on health would be more powerfully explicated by family functioning than by social support. We mailed a pilot survey to a random sample of patients attending a family practice center, including questions on depressive symptoms, cardiovascular health behaviors, demographics, social support using the ISEL scale, and family functioning using the FEICS scale. FEICS is a self-report questionnaire designed to assess family emotional involvement and criticism, the media elements of family expressed emotion. Eighty-three useable responses were obtained. Regression analyses and structural modelling showed both direct and indirect statistically significant paths from social relationships to health behaviors. Family criticism was directly associated (standardized coefficient = 0.29) with depressive symptoms, and family emotional involvement was directly associated with both depressive symptoms (coefficient = 0.35) and healthy cardiovascular behaviors (coefficient = 0.32). The results support the primacy of family functioning factors in understanding the associations among social relationships, mental health, and health behaviors. The contrasting relationships between emotional involvement and depressive symptoms on the one hand and emotional involvement and health behaviors on the other suggest the need for a more complex model to understand the connections between social relationships and health.