Social support and depression have been shown to affect the prognosis of coronary patients, and social support has been found to influence depression in community and patient samples. We investigated the characteristics of coronary patients whose depressive symptomatology was most likely to improve with social support. We predicted that social support would be most beneficial for the most severely depressed, the old, the poor, the most severely ill, and those with poor functional status. Patients (n = 590) with documented coronary artery disease were assessed for depressive symptoms, social support, and functional status while in hospital. They were reassessed for depression 1 month later during a home visit. Depression scores were lower at follow-up (p = 0.001), and improvement was more marked among those reporting more support (p <0.001). The social support effect was strongest among those with high levels of depression at baseline (p <0.001) and those with lower income (p = 0.01). Unexpectedly, social support was more strongly associated with improvement in younger patients (p = 0.01). Social support did not interact with gender, disease severity, or functional status. These findings are partially consistent with the notion that social support is most effective for those who are most vulnerable and/or have few coping resources. These findings also have implications for the design and interpretation of psychosocial interventions.