Objective: To investigate the effect of different levels of exercise on telomere length. Methods: CINAHL, SPORTDiscus (EBSCO), OVID (Medline) and EMBASE databases were searched for eligible studies. Methodological quality was evaluated using the Newcastle-Ottawa Scale, and heterogeneity among the studies was assessed using the I-squared test. When heterogeneity among studies was high (I2 > 50%), a random-effects model was used (Review Manager version 5, Cochrane Collaboration, Copenhagen, Denmark); otherwise, a fixed-effects model was used. Results: Eleven eligible studies involving 19,292 participants were included in this meta-analysis. Longer telomere length was associated with physically active individuals, with a mean difference (MD) of 0.15 (95% confidence interval; 95% CI 0.05, 0.24); I2 = 99%. Longer telomere length was significantly associated with robust exercise (MD 0.08 (95% CI 0.04, 0.12)); I2 = 99%, as was moderate exercise (MD 0.07 (95% CI 0.03, 0.11)); I2 = 100%. Subgroup analysis revealed that longer telomere length was positively associated with exercise, regardless of sex, but was not statistically significant in elderly populations. Conclusion: Compared with inactive individuals, telomere lengths were longer in active subjects, regardless of the intensity of exercise. Key words: exercise; telomere length; meta-analysis.