Previous meta-analyses have shown that high-intensity interval training (HIIT) is more suitable than moderate continuous training (MCT) for improving peak oxygen uptake (VO2peak) in patients with coronary artery disease. However, none of these meta-analyses have tried to explain the heterogeneity of the empirical studies in optimizing cardiac rehabilitation programs. Therefore, our aims were (a) to estimate the effect of MCT and HIIT on VO2peak, and (b) to find the potential moderator variables. A search was conducted in PubMed, Scopus, and ScienceDirect. Out of the 3,110 references retrieved, 29 studies fulfilled the selection criteria to be included in our meta-analysis. The mean difference was used as the effect size index. Our results showed significant enhancements in VO2peak after cardiac rehabilitation based on MCT and HIIT (mean difference = 3.23; 95% confidence interval [2.81, 3.65] ml·kg−1·min−1 and mean difference = 4.61; 95% confidence interval [4.02, 5.19] ml·kg−1·min−1, respectively), with greater increases after HIIT (p < .001). Heterogeneity analyses reached statistical significance with moderate heterogeneity for MCT (p < .001; I
2 = 67.0%), whereas no heterogeneity was found for the effect of HIIT (p = .220; I
2 = 22.0%). Subgroup analyses showed significant between-group heterogeneity of the MCT-induced effect based on the training mode (p < .001; I
2 = 90.4%), the risk of a new event (p = .010; I
2 = 77.4%), the type of cardiovascular event (p = .009; I
2 = 84.8%), the wait time to start cardiac rehabilitation (p = .010; I
2 = 76.6%), and participant allocation (p = .002; I
2 = 89.9%). Meta-regressions revealed that the percentages of patients undergoing a revascularization procedure (B = −0.022; p = .041) and cardiorespiratory fitness at baseline (B = −0.103; p = .025) were inversely related to the MCT-induced effect on the VO2peak.