To assess the use of glycosylated haemoglobin to discriminate between various degrees of glucose intolerance, glycosylated haemoglobin levels were determined in 107 subjects (48 males and 59 females, age range 18-80 years). Following a 75 g oral glucose tolerance test and according to World Health Organization criteria, subjects were classified as normal (n = 32), diabetic (n = 46) or as having impaired glucose tolerance (n = 29). Mean glycosylated haemoglobin levels were 5.8 +/- 1.3% (range 4%-9%) in normal subjects, 7.1 +/- 1.7% in subjects with impaired glucose tolerance (range 4.1%-10.1%) and 10.1 +/- 2.6% (range 4.7%-18.8%) in diabetic patients. The difference between the groups was highly significant (p less than 0.01). Twelve per cent of normal subjects exceeded and 52% of subjects with impaired glucose tolerance fell below 7.4% (mean +/- 2SD, considered as the upper limit of normal values). A significant correlation was observed between glycosylated haemoglobin values and fasting blood glucose (r = 0.68, p less than 0.01). These results provide evidence that glycosylated haemoglobin levels are influenced by slightly reduced carbohydrate tolerance. Glycosylated haemoglobin may be a useful test to improve the specificity of the oral glucose load to select and to follow-up subjects with impaired glucose tolerance.