The current classification and diagnostic criteria for diabetes mellitus were introduced by the United States National Data Group in 1979 and endorsed by the World Health Organization in 1980, with modifications in 1985 and 1994. The criteria, chosen to reflect the risk of complications, were the synthesis of considerable thought and expertise and represented a consensus which, it was hoped, would prove helpful to all those involved with diabetes practising clinician, research scientist and epidemiologist alike. The inconvenience, variability and nonphysiological nature of the oral glucose tolerance test (OGTT) are well-recognised. In spite of these limitations the 2-h post-load plasma glucose has remained the standard against which all other tests have been evaluated. This article reviews the original justification for the OGTT, and in the light of more recent epidemiological research seeks to place the current diagnostic criteria for diabetes into a pathophysiological, diagnostic and prognostic perspective.