Although epidemiologic studies carried out in Taiwan, Bangladesh, and Sweden have demonstrated a diabetogenic effect of arsenic, the mechanisms remain unclear and require further investigation. This paper reviewed the potential biological mechanisms of arsenic-induced diabetes mellitus based on the current knowledge of the biochemical properties of arsenic. Arsenate can substitute phosphate in the formation of adenosine triphosphate (ATP) and other phosphate intermediates involved in glucose metabolism, which could theoretically slow down the normal metabolism of glucose, interrupt the production of energy, and interfere with the ATP-dependent insulin secretion. However, the concentration of arsenate required for such reaction is high and not physiologically relevant, and these effects may only happen in acute intoxication and may not be effective in subjects chronically exposed to low-dose arsenic. On the other hand, arsenite has high affinity for sulfhydryl groups and thus can form covalent bonds with the disulfide bridges in the molecules of insulin, insulin receptors, glucose transporters (GLUTs), and enzymes involved in glucose metabolism (e.g., pyruvate dehydrogenase and α-ketoglutarate dehydrogenase). As a result, the normal functions of these molecules can be hampered. However, a direct effect on these molecules caused by arsenite at physiologically relevant concentrations seems unlikely. Recent evidence has shown that treatment of arsenite at lower and physiologically relevant concentrations can stimulate glucose transport, in contrary to an inhibitory effect exerted by phenylarsine oxide (PAO) or by higher doses of arsenite. Induction of oxidative stress and interferences in signal transduction or gene expression by arsenic or by its methylated metabolites are the most possible causes to arsenic-induced diabetes mellitus through mechanisms of induction of insulin resistance and β cell dysfunction. Recent studies have shown that, in subjects with chronic arsenic exposure, oxidative stress is increased and the expression of tumor necrosis factor α (TNFα) and interleukin-6 (IL-6) is upregulated. Both of these two cytokines have been well known for their effect on the induction of insulin resistance. Arsenite at physiologically relevant concentration also shows inhibitory effect on the expression of peroxisome proliferator-activated receptor γ (PPARγ), a nuclear hormone receptor important for activating insulin action. Oxidative stress has been suggested as a major pathogenic link to both insulin resistance and β cell dysfunction through mechanisms involving activation of nuclear factor-κB (NF-κB), which is also activated by low levels of arsenic. Although without supportive data, superoxide production induced by arsenic exposure can theoretically impair insulin secretion by interaction with uncoupling protein 2 (UCP2), and oxidative stress can also cause amyloid formation in the pancreas, which could progressively destroy the insulin-secreting β cells. Individual susceptibility with respect to genetics, nutritional status, health status, detoxification capability, interactions with other trace elements, and the existence of other well-recognized risk factors of diabetes mellitus can influence the toxicity of arsenic on organs involved in glucose metabolism and determine the progression of insulin resistance and impaired insulin secretion to a status of persistent hyperglycemia or diabetes mellitus. In conclusions, insulin resistance and β cell dysfunction can be induced by chronic arsenic exposure. These defects may be responsible for arsenic-induced diabetes mellitus, but investigations are required to test this hypothesis.