The worldwide prevalence of diabetes mellitus is expected to rise to over 550 million by 2030; this continuous increase being driven by aging and obesity [1]. Diabetes has direct human and monetary costs, but the main burden for patients and health care systems comes from vascular complications of diabetes. If diabetes is well recognized as a risk factor for myocardial infarction, stroke, and peripheral artery disease, it has also to be seen as a major provider of end-stage renal disease (ESRD) [2]. High plasma vasopressin concentrations have been consistently reported in experimental models of diabetes and patients with diabetes, either of type 1 or type 2 [3,4,5]. Interestingly, this was described in patients with uncontrolled diabetes, but also in those with a good glycemic control. However, the study of large populations was problematic due to the low stability of vasopressin, a short peptide, and the limited availability of high-quality assays. A few years ago, an assay for copeptin, the stable C-terminal portion of the precursor of vasopressin and as such a good surrogate for the secretion of vasopressin, was made commercially available [6]. Since then, nearly 300 papers based on copeptin measurements were published. Several works not only confirmed the elevation of copeptin in diabetic patients, but also extended this observation. Actually, plasma copeptin was already elevated in patients with metabolic abnormalities such as those preceding type 2 diabetes (metabolic syndrome) [7]. It has been validated as a prognostic marker of diabetes, independent of other classical risk factors like BMI or age. Water intake is a major determinant of vasopressin secretion. In the French D.E.S.I.R. study, a cohort recruited in the population, our group found that the risk of new-onset hyperglycemia during a 9-year follow-up was inversely and independently associated with daily water intake [8]. After adjustment for classical confounders and intake of alcoholic beverages and sweet drinks, participants who reported drinking more than 1 l water per day were 21% less likely to develop hyperglycemia than those with a water intake below 0.5 l/day (p = 0.016). Several lines of evidence support a causal role of vasopressin in these associations. First, in the liver, vasopressin stimulates glycogenolysis and gluconeogenesis through V1a receptor in vitro [9,10,11]. V1b, another G-protein-coupled vasopressin receptor, is expressed in pancreatic Langerhans islets, and perfused pancreas experiments revealed a role of vasopressin in glucagon and insulin secretion [12]. In healthy humans, acute infusion of vasopressin increases plasma glucose, due to an increased glucose production, which is consistent with the expected vasopressin action on the liver [13]. Finally, genetic association studies found a link between vasopressin receptor polymorphisms and metabolic disorders [14].