BACKGROUND: Despite many epidemiological studies on the relationship between social support during pregnancy and perinatal/postpartum depression, its impact remains unclear. Therefore, this study examined this association using a causal model of risk based on various levels of social support. METHODS: Participants were 88,711 mothers in an ongoing nationwide birth cohort study in Japan. Social support during pregnancy was set at four treatment levels. Depressive state was identified twice, with the Kessler Psychological Distress Scale (K6; score ≥ 5) and the Edinburgh Postnatal Depression Scale (EPDS; score ≥ 9). With the highest level of social support set as reference, marginal structural models were fitted to derive counterfactual risk ratios (cRRs). RESULTS: cRRs (95% CIs) for EPDS cases increased as the level of social support decreased, with values of 1.06 (0.99-1.13) for upper-middle, 1.30 (1.23-1.39) for lower-middle, and 1.61 (1.52-1.71) for low. Also, cRRs (95% CIs) at the second measurement (interaction) tended to increase as social support decreased, with values of 1.05 (0.97-1.13) for upper-middle, 1.05 (0.98-1.13) for lower-middle, and 1.10 (1.03-1.18) for low. The same tendency was observed in K6 cases. LIMITATION: Depressive state was identified by self-administered questionnaire. CONCLUSIONS: Lower social support during pregnancy is associated with increased counterfactual risk of perinatal/postpartum depressive state and worsened counterfactual spontaneous recovery over time. These findings highlight the importance of sufficient social support for preventing perinatal/postpartum depression and promoting recovery. TRIAL REGISTRATION: UMIN000030786.