The borderline between diabetes and intolerance to carbohydrates has been drawn on the basis of prospective studies which determined a glycaemic threshold marking the risk for microangiopathy. On the other hand, the borderline between intolerance to carbohydrates and normal glucose tolerance remains arbitrary: 25% for subjects who are intolerant to carbohydrates return to normal glucose tolerance within 10 years. This is due to the fact that intolerance to carbohydrates is a heterogeneous entity which should be dismembered according to the severity of insulin deficiency and to the degree of insulin resistance. Alteration of insulin secretion is perhaps the most specific marker of susceptibility to non insulin dependent diabetes, but insulin resistance is certainly the principal factor exhausting insulin secretion and leading to non insulin dependent diabetes. Insulin resistance and the hyperinsulinism it creates seem to facilitate atherogenesis, even when glucose tolerance is still normal, so that the oral glucose tolerance test is not only poorly reproducible but loses a great deal of its value in the early detection of vascular risk. Measurements of fasting and post-prandial glucose levels and of A1C haemoglobin, cholesterol, triglyceride, and HDL cholesterol levels usually make it possible to classify subjects into one of the three following categories: (1) no risk of macro- or microangiopathy; (2) diabetes with a risk of macro- or microangiopathy; (3) intolerance to glucose with risk of atherogenesis but no risk of microangiopathy. The oral glucose tolerance test probably remains useful within a small set of values that are either very slightly above normal or dissociated. Measuring blood insulin levels might be a better way of assessing the risk of atherogenesis, but the clinical use of this test requires evaluation.