ABSTRACT Purpose Postpartum depression (PPD) impacts about one out of eight new mothers. Research has demonstrated that social support is a protector of PPD. Nevertheless, there has been disagreement on how social support influences depression. The objective is to test two theories – main-effect theory and stress-buffering theory of social support on PPD with different definitions of stress and two types of social support. Methods Secondary longitudinal data from the National Data Archive on Child Abuse and Neglect were used. Parenting stress and difficult life circumstances (DLC) measured at six-month postpartum was used to predict the changes in depression from six- to 12-month postpartum; social support at six-month postpartum was conceptualized as a moderator between stress and PPD. Structural Equation Modeling was adopted for the longitudinal analyses. Results DLC and parental distress both showed long-lasting impacts on PPD. Social support was not found to have a direct or indirect effect on PPD. Neither the stress-buffering theory nor the main-effect model was endorsed by this study. Conclusion Interventions that are accessible to new mothers and aim to improve self-efficacy are recommended. Measures with higher psychometric quality should be used in moderation research. More longitudinal studies with shorter lags between measurement occasions are warranted.