Background: Dietary and metabolic factors have been associated with the risk of Barrett's esophagus (BE) in observational epidemiological studies. However, the aforementioned associations may be influenced by confounding bias. The present study aimed to evaluate these causal relationships through univariate and multivariate Mendelian randomization (MR) analysis. Methods: Genetic instruments associated with dietary and metabolic factors were obtained in the large-scale genome-wide association studies (GWAS), respectively. Summary data for BE were available from a GWAS of 13,358 cases and 43,071 controls of European descent. Univariable MR analysis was initially performed to estimate the causal relationship between exposures and BE. The inverse-variance weighted (IVW) method was adopted as the primary MR analysis. Multivariate MR analysis was further conducted to evaluate the independent effects of exposures on BE. Results: In univariate MR analysis, BE was causally associated with higher body mass index (odds ratio (OR) = 2.575, 95% confidence interval (CI): 2.301-2.880, P = 7.369E-61), larger waist circumference (OR = 2.028, 95% CI: 1.648-2.496, P = 2.482E-11), and smoking per day (OR = 1.241, 95% CI: 1.085-1.419, P = 0.002). Dried fruit intake showed a protective effect on BE (OR = 0.228, 95% CI: 0.135-0.384, P = 2.783E-08), whereas alcohol drinking, coffee intake, tea intake, fresh fruit intake, and type 2 diabetes mellitus were not associated with BE (P = 0.351, P = 0.458, P = 0.125, P = 0.847, P = 0.413, respectively). No pleiotropy was found in the sensitivity analysis. The relationships of obesity, smoking, and dried fruit intake with BE risk remained strong after adjustment. Conclusions: Our study provided MR evidence supporting obesity and smoking were independent risk factors for BE. Conversely, dried fruit intake was a protective factor for BE.