Purpose This study evaluated the clinical efficacy of continuous passive motion (CPM) following knee arthroplasty and determined the predictors of effect sizes of range of motion (ROM) and functional outcomes in patients with knee arthritis. Methods A comprehensive electronic database search was performed for randomized controlled trials (RCTs), without publication year or language restrictions. The included RCTs were analyzed through meta-analysis and risk of bias assessment. Study methodological quality (MQ) was assessed using the Physiotherapy Evidence Database (PEDro) scale. Inverse-variance weighted univariate and multivariate metaregression analyses were performed to determine the predictors of treatment outcomes. Results A total of 77 RCTs with PEDro scores ranging from 6/10 to 8/10 were included. Meta-analyses revealed an overall significant favorable effect of CPM on treatment success rates [odds ratio: 3.64, 95% confidence interval (CI) 2.21-6.00]. Significant immediate [postoperative day 14; standard mean difference (SMD): 1.06; 95% CI 0.61-1.51] and short-term (3-month follow-up; SMD: 0.80; 95% CI 0.45-1.15) effects on knee ROM and a long-term effect on function (12-month follow-up; SMD: 1.08; 95% CI 0.28-1.89) were observed. The preoperative ROM, postoperative day of CPM initiation, daily ROM increment, and total application days were significant independent predictors of CPM efficacy. Conclusion Early CPM initiation with rapid progress over a long duration of CPM application predicts higher treatment effect on knee ROM and function. The results were based on a moderate level of evidence, with good MQ and potential blinding biases in the included RCTs. An aggressive protocol of CPM has clinically relevant beneficial short-term and long-term effects on postoperative outcomes. Level of evidence II.