The revision of the classification of diabetes mellitus, to differentiate clearly between insulin-dependent (IDDM) and non-insulin-dependent diabetes mellitus (NIDDM), and the provision of unambiguous guidelines for diagnosis (1) constitute important recent developments in diabetes epidemiology. However, our knowledge even of the prevalence of NIDDM remains surprisingly incomplete for many areas of the world. Whilst NIDDM may still be uncommon in rural Africa, prevalence has been reported as approximately 10% in blacks in the United States, indicating a need for more information regarding the prevalence of NIDDM in urban Africa. There is also little information with regard to urban communities in Latin America. The highest prevalence of NIDDM is found in certain indigenous North American and Western Pacific societies. In extreme cases approximately one-third of the adult population now suffers from the disease. NIDDM is also common (prevalence approximately 5%) in Europe and in communities of European origin. Data from the United States suggest that approximately one-fifth of white North Americans can expect to develop NIDDM if they live to the seventh decade of life. Of populations of South-East Asian ethnicity, Indians appear to be the most susceptible. Indian migrants to Fiji, South Africa and South America all demonstrate prevalence of NIDDM of 10% or more. Whilst it was formerly believed that Chinese were rarely affected by NIDDM, recent reports cast some doubt on this. Prevalence of NIDDM in the Western Pacific varies widely. However, with the possible exception of certain Melanesian populations, prevalence is high in all communities which have abandoned their traditional lifestyle.(ABSTRACT TRUNCATED AT 250 WORDS)