Exercise is a form of physical activity (PA). PA is an important marker of health and quality of life in older adults. The purpose of this study was to conduct a systematic review of the literature to assess the effect of exercise-based interventions on an at least six-month follow up PA measure, and to describe the specific strategies implemented during the intervention to strengthen the sustainability of PA in community-dwelling 65+ year-old adults. We registered and conducted a systematic review and meta-analysis (PROSPERO: CRD42017070892) of randomized clinical trials (RCT). We searched three electronic databases during January 2018 to identify RCT assessing any type of exercise-based intervention. Studies had to report a pre-, post-, and at least 6-month post-intervention follow-up. To be included, at least one PA outcome had to be assessed. The effect of exercise-based interventions was assessed compared to active (e.g., a low-intensity type of exercise, such as stretching or toning activities) and non-active (e.g., usual care) control interventions at several time points. Secondary analyses were conducted, restricted to studies that reported specific strategies to enhance the sustainability of PA. The intervention effect was measured on self-reported and objective measures of time spent in PA, by means of standardized mean differences. Standardized mean differences of PA level were pooled. Pooled estimates of effect were computed with the DerSimonian–Laird method, applying a random effects model. The risk of bias was also assessed. We included 12 studies, comparing 18 exercise intervention groups to four active and nine non-active control groups. Nine studies reported specific strategies to enhance the long-term sustainability of PA. The strategies were mostly related to the self-efficacy, self-control, and behavior capability principles based on the social cognitive theory. Exercise interventions compared to active control showed inconclusive and heterogeneous results. When compared to non-active control, exercise interventions improved PA time at the six-months follow up (standardized mean difference (SMD) 0.30; 95%CI 0.15 to 0.44; four studies; 724 participants; I2 0%), but not at the one- or two-years follow-ups. No data were available on the mid- and long-term effect of adding strategies to enhance the sustainability of PA. Exercise interventions have small clinical benefits on PA levels in community-dwelling older adults, with a decline in the observed improvement after six months of the intervention cessation.