Randomized controlled trials (RCTs) from multiple databases from inception to December 2020 published in English were searched. Two researchers independently performed eligible study screening and data extraction. The methodological quality was assessed with the Jadad score. Meta-analysis of RCTs on TC in the treatment of MCI was performed with RevMan Version 5.4.1. Seven RCTs with 1265 participants were included. For most RCTs, the overall reporting of methodological quality was high. Results of the meta-analysis indicate that TC improved MCI patients' cognitive function significantly, including overall cognitive function (MD = -2.24, 95% CI -3.51 to -0.97, P = 0.0005), memory and learning (SMD = 0.83, 95% CI 0.22 to 1.45, P = 0.008), visuospatial ability (MD = 3.15, 95% CI 0.74 to 5.56, P = 0.01), executive functions (MD = 0.32, 95% CI 0.03 to 0.61, P = 0.03), and physical activity (MD = 18.78, 95% CI 10.80 to 26.76, P < 0.00001). However, no significant benefit was found for TC on psychological activity (MD = 0.17, 95% CI -0.62 to 0.96, P = 0.36) and biomarker improvement. The meta-analysis confirmed the clinical therapeutic effect of TC for MCI. More rigorous and long-term follow-up RCTs should be conducted in the future. Copyright © 2021 Run Lin et al.