The American Diabetes Association have recommended that the fasting plasma glucose level for the diagnosis of diabetes is lowered and that this becomes the main diagnostic test. We have used population-based data from three ethnic groups in Newcastle upon Tyne to examine the implications of this change. Data were available on 824 European (25-74 years), 375 Chinese (25-64 years), and 680 South Asian (25-74 years) subjects. All subjects apart from those reporting a prior diagnosis of diabetes underwent a standard 75 g oral glucose tolerance test (WHO criteria) which included the measurement of fasting glucose. The prevalence of diabetes was higher in all three ethnic groups using the new ADA criteria compared to the WHO criteria: 7.1% vs 4.8% in Europeans; 6.2% vs 4.7% in Chinese; and 21.4% vs 20.1% in South Asians. There was much variation in individuals categorized by the ADA and WHO criteria. Agreement between the two for the diagnosis of previously unknown diabetes was only moderate (kappa statistics 0.42 to 0.59). Thus in the populations studied the new criteria would increase the prevalence of diabetes in addition to classifying some individuals diabetic by current criteria as non-diabetic. It should be stressed however that diagnosis of the individual should not be based on a single test.