Chronic arsenic exposure has been suggested to contribute to diabetes development. We performed a systematic review of the experimental and epidemiologic evidence on the association of arsenic and type 2 diabetes. We identified 19 in vitro studies of arsenic and glucose metabolism. Five studies reported that arsenic interfered with transcription factors involved in insulin-related gene expression: upstream factor 1 in pancreatic beta-cells and peroxisome proliferative-activated receptor gamma in preadipocytes. Other in vitro studies assessed the effect of arsenic on glucose uptake, typically using very high concentrations of arsenite or arsenate. These studies provide limited insight on potential mechanisms. We identified 10 in vivo studies in animals. These studies showed inconsistent effects of arsenic on glucose metabolism. Finally, we identified 19 epidemiologic studies (6 in high-arsenic areas in Taiwan and Bangladesh, 9 in occupational populations, and 4 in other populations). In studies from Taiwan and Bangladesh, the pooled relative risk estimate for diabetes comparing extreme arsenic exposure categories was 2.52 (95% confidence interval, 1.69-3.75), although methodologic problems limit the interpretation of the association. The evidence from occupational studies and from general populations other than Taiwan or Bangladesh was inconsistent. In summary, the current available evidence is inadequate to establish a causal role of arsenic in diabetes. Because arsenic exposure is widespread and diabetes prevalence is reaching epidemic proportions, experimental studies using arsenic concentrations relevant to human exposure and prospective epidemiologic studies measuring arsenic biomarkers and appropriately assessing diabetes should be a research priority.