Studies on the association between arsenic exposure and diabetes mellitus (DM) yielded inconsistent results. Epidemiologic data on the associations between arsenic exposures via inhalation and DM are limited. Therefore, we conducted a meta-analysis to evaluate the risk of DM associated with arsenic exposure. We searched the related literature through a systematic approach and analyzed the data according to the exposure route (inhalation and ingestion). We used random-effect models to estimate the summary relative risks (RRs) for DM associated with arsenic exposure and used I2 statistics to assess the heterogeneity of studies. We identified 38 relevant studies, of which the 32 on the ingestion route showed a significant association between arsenic exposure and DM (RR = 1.57; 95% CI 1.27–1.93). Focusing on the 24 studies in which the diagnosis of DM was confirmed using laboratory tests or medical records, we found that the summary RR was 1.71 (95% CI 1.32–2.23), very close to the overall estimates. We concluded that ingested arsenic is associated with the development of DM, but the heterogeneity among the studies may affect the results.