The Islington Diabetes Survey identified two groups of non-diabetic individuals, low and high glycators, who remained consistently classified 4.4 +/- 0.2 years after the original study. To investigate the mechanism for this grouping, 12 original subjects, 5 with low and 7 with high levels of glycated haemoglobin relative to their 2 h blood glucose, were studied. Glycated albumin and fructosamine measurements gave comparable classifications, with three individuals being misclassified for each measurement; in addition glycated albumin was positively correlated with mean blood-glucose concentration (r = 0.53; P < 0.05). Fasting plasma glucose concentration was greater than the intra-erythrocyte concentration (P < 0.05), but their ratio was reduced in low compared to high glycators (0.77 +/- 0.12 and 0.94 +/- 0.13, P < 0.0001). No differences between groups were found for plasma insulin, urea or non-esterified fatty acids; plasma or intra-erythrocyte inorganic phosphate or vitamin C; nor plasma, erythrocyte or urinary total amino acids. Erythrocyte 2,3-diphosphoglycerate, a catalyst of glycation, was elevated in high compared to low glycators (5.61 +/- 0.26 and 4.81 +/- 0.24 mmol/l, P < 0.001). Mean centile glycated haemoglobin was positively correlated with intra-erythrocyte pH (r = 0.55; P < 0.05) and negatively with plasma total amino acids (r = -0.57, P < 0.05). These data indicate that the intra-erythrocyte environment of high glycators favours glycation of haemoglobin. This could have important consequences for diabetic patients in terms of monitoring their glycaemic control and in the progression of those complications related to non-enzymic glycation of intracellular proteins.