The Thinking Healthy Program (THP) is a multicomponent low-intensity cognitive behavioral therapy-based psychosocial intervention. This intervention has been shown to be clinically effective in perinatal depression (PND) and feasible for implementation in low-resourced settings. It has also been shown to work universally for different phenotypes of PND. However, the mechanism through which THP resolves different phenotypes of PND are unclear. The present investigation presents secondary mediation analyses of a dataset curated from a cluster randomized controlled trial conducted in Pakistan assessing the effectiveness of the THP. Women aged 16–45 years in their third pregnancy trimester, with a diagnosis of PND, underwent 16 sessions of the intervention. The severity of depression was assessed using the Hamilton Depression Rating Scale (HDRS). 2-1-1 mediation models revealed that social support exerted significant mediation in driving the intervention effects for improving the severity of depressive symptoms on the HDRS [B (SE) = 0.45 (0.09), 95% CI: 0.30–0.60] and its symptom dimensions of core emotional symptoms [B (SE) = 0.27 (0.06), 95% CI: 0.18–0.37], somatic symptoms [B (SE) = 0.24 (0.04), 95% CI: 0.16–0.31] and insomnia symptoms [B (SE) = 0.04 (0.02), 95% CI: 0.02–0.07].