Exercise could alleviate non-alcoholic fatty liver disease (NAFLD), but it was not clear which exercise methods could effectively treat NAFLD. The purpose of this systematic review and meta-analysis was to evaluate the effects of different exercise patterns on eight indicators in patients with NAFLD. We searched PubMed, Scopus, Web of Science, China National Knowledge Infrastructure, and the Wanfang Data from its inception to 30 June 2020. This review includes all randomized controlled trials (RCT) that assessed and compared the effects of different exercise on eight indicator parameters in patients with NAFLD. The results indicate that aerobic exercises could significantly improve the eight indicators in patients with NAFLD including triglycerides (TG, weighted mean difference (WMD) = -0.53, 95%CI: -0.68~-0.39, Z = 7.37, p < 0.01), total cholesterol (TC, WMD = -0.39, 95%CI: -0.55~-0.23, Z = 4.76, p < 0.01), low density lipoprotein (LDL, WMD = -0.47, 95%CI: -0.68~-0.26, Z = 4.33, p < 0.01), high density lipoprotein (HDL, WMD = 0.12, 95%CI: 0.05~0.18, Z = 3.56, p < 0.01), alanine aminotransferase (ALT, WMD = -6.14, 95%CI: -10.99~-1.29, Z = 2.48, p < 0.05), aspartate aminotransferase (AST, WMD = -5.73, 95%CI: -9.08~-2.38, Z = 3.36, p < 0.01), and body mass index (BMI, WMD = -0.85, 95%CI: -1.19~-0.51, Z = 4.92, p < 0.01). Resistance exercises could significantly reduce the levels of TG (WMD = -0.56, 95%CI: -0.85~-0.28, Z = 3.86, p < 0.01) and AST (WMD = -2.58, 95%CI: -4.79~-0.36, Z = 2.28, p < 0.05) in the patients. High-intensity interval training could significantly improve the level of ALT (WMD = -6.20, 95%CI: -9.34~-3.06, Z = 3.87, p < 0.01) in patients with NAFLD. These three exercise methods had different effects on the eight indexes of NAFLD in our present meta-analysis, providing some reference for the establishment of exercise prescription for patients with NAFLD. Keywords: exercises; meta-analysis; non-alcoholic fatty liver disease; randomized clinical trials.