This study was performed to identify the association between smoking and Alzheimer's disease (AD). To perform this meta-analysis based on case-control and cohort studies, PubMed, Google Scholar, and the CNKI electronic databases were searched through April 30, 2017. Our meta-analysis included 27 studies, including 16 that reported odds ratios (ORs) and 11 that reported hazard ratios (HRs) or ratio risks. No significant association was found between smoking and AD among the studies that reported ORs (1.020, 95% confidence interval [CI] = 0.812-1.281, I = 67.9%, random model, p < 0.001). A subgroup analysis revealed no significant difference between different smoking statuses. The pooled HRs revealed a significant association between smoking and AD (HR = 1.520, 95% CI = 1.194-1.934, I = 83.6%, random model, p < 0.001). Cumulative meta-analysis of the HRs revealed that the effect of smoking on AD tended to be stable over time. Smoking may confer an increased risk of AD, and this effect has tended to be stable over time.