Background: Cognitive behavioral therapy (CBT) has been widely used for post-stroke depression (PSD), but the findings have been inconsistent. This is a meta-analysis of randomized controlled trials (RCTs) of CBT for PSD. Methods: Both English (PubMed, PsycINFO, Embase) and Chinese (WanFang Database, Chinese National Knowledge Infrastructure and SinoMed) databases were systematically searched. Weighted and standardized mean differences (WMDs/SMDs), and the risk ratio (RR) with their 95% confidence intervals (CIs) were calculated using the random effects model. Results: Altogether 23 studies with 1,972 participants with PSD were included and analyzed. Of the 23 RCTs, 39.1% (9/23) were rated as high quality studies, while 60.9% (14/23) were rated as low quality. CBT showed positive effects on PSD compared to control groups (23 arms, SMD = −0.83, 95% CI: −1.05 to −0.60, P < 0.001). Both CBT alone (7 arms, SMD = −0.76, 95% CI: −1.22 to −0.29, P = 0.001) and CBT with antidepressants (14 arms, SMD = −0.95, 95% CI: −1.20 to −0.71, P < 0.00001) significantly improved depressive symptoms in PSD. CBT had significantly higher remission (6 arms, RR = 1.76, 95% CI: 1.37–2.25, P < 0.00001) and response rates (6 arms, RR = 1.41, 95% CI: 1.22–1.63, P < 0.00001), with improvement in anxiety, neurological functional deficits and activities of daily living. CBT effects were associated with sample size, mean age, proportion of male subjects, baseline depression score, mean CBT duration, mean number of CBT sessions, treatment duration in each session and study quality. Conclusion: Although this meta-analysis found positive effects of CBT on depressive symptoms in PSD, the evidence for CBT is still inconclusive due to the limitations of the included studies. Future high-quality RCTs are needed to confirm the benefits of CBT in PSD.