Background Elevated levels of fetuin-A are associated with increased risks of metabolic syndrome, type 2 diabetes and nonalcoholic fatty liver disease. This meta-analysis investigated whether exercise interventions can reduce fetuin-A in adults. Methods We searched clinical trials that objectively assessed fetuin-A and included study arms with exercise intervention. The pre-intervention and post-intervention data were used for meta-analysis. The effect sizes were calculated as standardized mean differences or changes in fetuin-A and expressed as Hedges’ g using random-effects models. Results The overall Hedges’ g for fetuin-A in all included interventions was − 0.640 (95%CI − 1.129 to − 0.151; n = 9), but this effect was not observed in obese (g = − 0.096; 95%CI, − 0.328 to 0.135) and type 2 diabetes/dysglycemia (g = − 0.56; 95%CI, − 1.348 to 0.236) individuals. Additionally, the random-effects meta-regression analysis showed that there was not a greater decrease in fetuin-A in individuals who achieved greater body mass index reductions (regression coefficient = 0.065; 95%CI, − 0.185 to 0.315). Conclusion Supervised exercise is associated with reductions in fetuin-A levels in adults and elderly. However, the results of the present meta-analysis should be interpreted with caution because of the variety of type of exercises and individual obesity related-disorders involve. Therefore, additional high-quality randomized controlled trials describing the effect of supervised exercise interventions on fetuin-A in adults are still needed.  Electronic supplementary material The online version of this article (10.1186/s12944-019-0962-2) contains supplementary material, which is available to authorized users.