Background: To assess the effectiveness of orthodontic temporary anchorage devices in canine retraction during two-step technique. Methods: The search was systematically performed up to June 2019 in five electronic databases (PubMed, Embase, Cochrane Central Register of Controlled Trials, Web of Science, Scopus). A risk of bias assessment was performed using the Cochrane Risk of Bias Tool for Randomized Controlled Trials (RCTs) and the Risk of Bias in Nonrandomized Studies - of Interventions (ROBINS-I) tool for controlled clinical trials (CCTs) . The Grades of Recommendation, Assessment, Development and Evaluation (GRADE) approach was applied for quality assessment. Data concerning the mean difference in mesial molar movement and extent of canine retraction were extracted for statistical analysis. The mean difference and 95% confidence interval were analyzed for continuous data. A meta-analysis with a random-effects model for comparable outcomes was carried out. Results: Three RCTs and five CCTs were finally included. Meta-analysis showed a significant increase not only in anchorage preservation in the implant anchorage group in both the maxilla (1. 56 mm , 95% CI: 1.14 to 1.98, P <0.00001 ) and the mandible (1.62 mm , 95% CI: 1.24 to 2.01, P <0.00001 ) but also in canine retraction in the implant anchorage group in both the maxilla (0.43 mm , 95% CI: 0.16 to 0.69, P =0.001 ) and the mandible (0.26 mm , 95% CI: 0.02 to 0.49, P =0.03 ). Conclusions: Very low-quality evidence shows that implant anchorage is more efficient than conventional anchorage during canine retraction. Further high-quality studies are needed.