Causal relationships between perinatal bonding failure, depression, and social support among mothers remain unclear. A total of 494 women (mean age 32.4 ± 4.5 years) completed the Mother-Infant Bonding Questionnaire (MIBQ), the Edinburgh Postnatal Depression Scale (EPDS), and the Japanese version of the Social Support Questionnaire in early pregnancy before week 25 (T1) and 1 month after delivery (T2). Our model of recursive structured equation modeling (SEM) showed acceptable fit (CMIN/df = 2.2, CFI = 0.97, and RMSEA = 0.05). It was revealed that: (1) a lower number of supportive persons at T1 significantly predicted both MIBQ and EPDS scores at T1 and T2; (2) at T1, poorer satisfaction with the social support received significantly predicted EPDS scores; (3) both MIBQ and EPDS scores at T1 significantly predicted their respective scores at T2. Out cohort study indicates that the number of individuals who are available to provide social support and the degree of satisfaction with the level of social support received during pregnancy have a great influence on bonding failure and depression in the postpartum period. These findings suggest that psychosocial interventions that focus on these two aspects of social support during pregnancy are effective in preventing bonding failure and depression in the postpartum period.