OBJECTIVES:To examine the effects of exercise training on cognitive function in individuals at risk of or diagnosed with Alzheimer's disease (AD). DESIGN: Meta-analysis. SETTING: PubMed, Scopus, ClinicalTrials.gov, and ProQuest were searched from inception until August 1, 2017. PARTICIPANTS: Nineteen studies with 23 interventions including 1,145 subjects with a mean age of 77.0 AE 7.5 were included. Most subjects were at risk of AD because they had mild cognitive impairment (64%) or a parent diagnosed with AD (1%), and 35% presented with AD. INTERVENTION: Controlled studies that included an exercise-only intervention and a nondiet, nonexercise control group and reported pre-and post-intervention cognitive function measurements. MEASUREMENTS: Cognitive function before and after the intervention and features of the exercise intervention. RESULTS: Exercise interventions were performed 3.4 AE 1.4 days per week at moderate intensity (3.7 AE 0.6 metabolic equivalents) for 45.2 AE 17.0 minutes per session for 18.6 AE 10.0 weeks and consisted primarily of aerobic exercise (65%). Overall, there was a modest favorable effect of exercise on cognitive function (d + = 0.47, 95% confidence interval (CI) = 0.26-0.68). Within-group analyses revealed that exercise improved cognitive function (d +w = 0.20, 95% CI = 0.11-0.28), whereas cognitive function declined in the control group (d +w = À0.18, 95% CI = À0.36 to 0.00). Aerobic exercise had a moderate favorable effect on cognitive function (d +w = 0.65, 95% CI = 0.35-0.95), but other exercise types did not (d +w = 0.19, 95% CI = À0.06-0.43).CONCLUSION: Our findings suggest that exercise training may delay the decline in cognitive function that occurs in individuals who are at risk of or have AD, with aerobic exercise possibly having the most favorable effect. Additional randomized controlled clinical trials that include objective measurements of cognitive function are needed to confirm our findings. J Am Geriatr Soc 0:1-9, 2018.