Frailty and depression were linked in observational studies, but the causality remains ambiguous. We intended to explore it using Mendelian randomization (MR). We obtained frailty genome-wide association study (GWAS) data from UK Biobank and TwinGen meta-analysis, and depression GWAS data from Psychiatric Genomics Consortium (PGC) and FinnGen (respectively recorded as PD and FD). We performed univariable and multivariable-adjusted MR with adjustments for body mass index (BMI) and physical activity (PA). Frailty was significantly associated with elevated risks of PD (OR, 1.860; 95% CI, 1.439 to 2.405; P < 0.001) and FD (OR, 1.745; 95% CI, 1.193 to 2.552; P = 0.004), and depression was meanwhile a susceptible factor for frailty (PD: β, 0.146; 95% CI, 0.086 to 0.201; P < 0.001; and FD: β, 0.112; 95% CI, 0.051 to 0.174; P < 0.001). This association was robust after adjustments for BMI or PA. Our study provides evidence of the bidirectional causal association between frailty and depression from genetic perspectives.