Several reviews focused on the use of digital cognitive behavioral therapy (CBT) across different populations. This review synthesized randomized controlled trials (RCTs) that evaluated the effectiveness of digital CBT on psychological symptoms (depression, anxiety, and stress symptoms). An extensive search was conducted in 10 databases from inception until August 29, 2021. A meta-analysis using a random-effects model was performed using Hedges' g. The potential sources of heterogeneity were explored through subgroup analyses and meta-regression analysis. A total of 18 RCTs in 2514 perinatal women were identified from over 23 countries. The sample size ranged from 25 to 910. Meta-analyses showed that digital CBT significantly reduced depression (g = -0.56, 95% CI: -0.85, -0.27, I2 = 84.81%, p < 0.001), anxiety (g = -0.30, 95% CI: -0.44, -0.17, I2 = 0%, p < 0.001), and stress (g = -0.75, 95% CI: -0.95, -0.56, I2 = 0%, p < 0.001) symptoms at post-intervention and stress symptoms at follow-up (g = -0.52, 95% CI: -0.93, -0.11, I2 = 0%, p = 0.01) compared with those in the control group. Subgroup analyses highlighted that the intervention was effective when CBT was combined with other therapeutic components which delivered via a mobile application. Preferable features of intervention should be more than eight sessions and conducted for than 6 weeks among postnatal women. Multivariable meta-regression showed that age was a significant covariate on depression symptoms. The sample size in the selected RCTs was small, and the overall quality of the evidence was very low. Digital CBT is a potential intervention for alleviating psychological outcomes in perinatal women. This review suggests the essential features to optimize the intervention effect. Further well-designed RCTs with large sample sizes are necessary.