According to previous epidemiological studies, we can reduce the thickness of epicardial fat and improve cardiovascular risk factors through exercise, and the changes may depend on the form of exercise. We systemically reviewed published studies that evaluated exercise intervention on epicardial adipose tissue (EAT) levels. We included randomized controlled trials (RCTs) comparing one exercise with another exercise or diet for the treatment to reduce EAT. We used fixed effects models for meta-analyses; effects of exercise on outcomes were described as mean differences (MD) or standardized difference of means (SMD) was used, their 95% confidence intervals (CI). Five RCTs were included (n = 299), 156 in exercise group and 143 in the control. In comparison to the control group, exercise significantly reduced EAT (SMD − 0.57, 95%CI − 0.97 to − 0.18) and waist circumference (MD − 2.95 cm, 95%CI − 4.93 to − 0.97). Exercise did not have an effect on BMI (MD − 0.23 kg/m2, 95%CI − 0.73 to 0.27), weight (MD − 0.06 kg, 95%CI − 1.46 to 1.34), or HDL (SMD 0.26, 95%CI − 0.06 to 0.57).VO2 was significantly increased by exercise (SMD 1.58, 95%CI 1.17 to 1.99). Risk of bias was high for 3 studies, and GRADE quality of evidence was very low to moderate. Exercise reduced epicardial adipose tissue and waist circumference, and did not have effect on weight, BMI, or HDL. Newer trials with better design and methods are necessary to improve the quality of the evidence. PROSPERO registration number (CRD42018096581)