Diabetes mellitus (DM) is associated with many infectious diseases, but the relationship between diabetes and urinary tract infections remains controversial. Therefore, we investigated, by Mendelian randomization (MR), the causal relationship between type 1 DM (T1DM), type 2 DM (T2DM), and 3 common urinary tract infections at different sites. We evaluated the causal relationship between T1DM and T2DM and pyelonephritis, cystitis, and urethritis in a two-sample MR study, using the inverse variance weighted (IVW) method for the primary analysis and MR-Egger and weighted median estimation methods as complementary methods. In addition, we conducted the heterogeneity test, the pleiotropy test, the MR pleiotropy residual sum and outlier test, and the leave-one-out sensitivity test to validate the reliability of the results. If the univariate analysis revealed a significant causal relationship, multivariate MR analysis was performed to adjust for genetically predicted body mass index and glycosylated hemoglobin. MR using IVW showed that patients with T1DM or T2DM were at increased risk of developing pyelonephritis. After adjusting for body mass index and glycosylated hemoglobin, respectively, T1DM and T2DM still had a direct causal relationship with pyelonephritis. Multivariate MR-IVW, MR-Egger, least absolute shrinkage and selection operator, and Robust all showed consistent causal relationships. However, no significant causal relationship was found between T1DM and T2DM and cystitis or urethritis. Our study revealed a causal relationship between T1DM and T2DM and the occurrence of pyelonephritis in individuals of European ancestry, but no significant causal association between T1DM and T2DM and cystitis or urethritis.