A model of maternal postpartum depression was tested in which difficult infant temperament was construed as a stressor and supportive interpersonal relationships were construed as a protective resource. It was hypothesized that both infant temperamental difficulty and level of social support would affect maternal depression through the cognitive mediation of perceived self-efficacy in the parenting role. Participants were 55 married women who were assessed during pregnancy and again 3 months postpartum. Infant temperament was assessed through observation, maternal crying records, and the Revised Infant Temperament Questionnaire. Results of a path analysis indicated that infant temperamental difficulty was strongly related to the mothers' level of postpartum depression, both directly and through the mediation of parenting self-efficacy. Consistent with predictions, social support appeared to exert its protective function against depression primarily through the mediation of self-efficacy. Both practical implications for identifying women at risk for postpartum depression and theoretical implications for understanding the mechanisms through which stressful events and social support affect adjustment are discussed.