Background: Alzheimer’s disease (AD), as the most common cause of dementia, brings huge economic burden for patients and social health care systems, which motivates researchers to study multiple protective factors, among which physical activity and exercise have been proven to be both effective and economically feasible. Methods: A systematic literature search was performed for eligible studies published up to November 1st 2018 on three international databases (PubMed, Cochrane Library, and Embase) and two Chinese databases (Wanfang Data, China National Knowledge Infrastructure). All analyses were conducted using Stata 14.0. Due to heterogeneity between studies, a random-effects model was used for this meta-analysis. Meta-analysis was used to explore if physical activity and exercise can exert positive effects on cognition of elderly with AD and subgroup analyses were conducted to find out if there are dose-response effects. Results: A total of 13 randomized controlled trials were included with a sample size of 673 subjects diagnosed with AD. Intervention groups showed a statistically significant improvement in cognition of included subjects measured by the MMSE score (SMD = 1.12 CI:0.66~1.59) compared to the control groups. Subgroup analyses showed different amounts of physical activity and exercise can generate different effects. Conclusions: As one of few meta-analyses comparing different quantities of physical activity and exercise interventions for AD in details, our study suggests that physical activity and exercise can improve cognition of older adults with AD. While the concomitant effects on cognition functions of high frequency interventions was not greater than that of low frequency interventions, the threshold remains to be settled. However, more RCTs with rigorous study design are needed to support our findings.