Are shorter telomeres causal risk factors for Alzheimer's disease (AD)? This study aimed to examine if shorter telomeres were causally associated with a higher risk of AD using Mendelian randomization (MR) analysis. Two-sample MR methods were applied to the summary effect sizes and standard errors from a genome-wide association study for AD. Twenty single nucleotide polymorphisms of genome-wide significance were selected as instrumental variables for leukocyte telomere length. The main analyses were performed primarily using the random-effects inverse-variance weighted method and complemented with the other three methods: weighted median approaches, MR-Egger regression, and weighted mode approach. The intercept of MR-Egger regression was used to assess horizontal pleiotropy. We found that longer telomeres were associated with lower risks of AD (odds ratio = 0.79, 95% confidence interval: 0.67, 0.93, P = 0.004). Comparable results were obtained using weighted median approaches, MR-Egger regression, and weighted mode approaches. The intercept of the MR-Egger regression was close to zero. This may show that there was not suggestive of horizontal pleiotropy. Our findings provided additional evidence regarding the putative causal association between shorter telomere length and the higher risk of AD.