From an urban population (n = 9,033) of 47-49-year-old males, 6,956 participated in a multiphase screening programme, of whom 1.5% were already registered as diabetic patients, 1.7% were then found to be diabetic; of 6,325 subjects given oral glucose tolerance tests, 6.6% were found to have impaired glucose tolerance (WHO criteria, 1985). In 889 asymptomatic cases with initial capillary whole blood glucose values greater than or equal to 6.6 mmol/l fasting and/or 2 h post-load, fluctuation in oral glucose tolerance was studied at repeat tests within one month; the mean differences in glucose values between the first and second test were less than -1% (fasting) and -15% (2 h post-load); there were no differences in body weight, and 62% of those with initially impaired glucose tolerance had normalised by the repeat test. Only in 109 cases (1.7%) were 2 h post-load values in the 7.8-11.0 mmol/l range both at the first and the repeat test; these cases were comparable vis-à-vis body mass index, triceps skin fold, blood pressure and initial glucose and insulin values, but had significantly lower oxygen uptake (2.34 +/- 0.54 l/min vs 2.63 +/- 0.68 l/min; p less than 0.003), as compared with subjects with initially impaired glucose tolerance but normal repeat test outcome. However, subjects with high normal first test results (2 h value in the 7.0-7.7 mmol/l range) and second test results in the 7.0-11.0 mmol/l range, resembled those with persistent impairment of glucose tolerance in all respects (including oxygen uptake). The repeat test procedure (including ergometry), is therefore to be recommended in selecting true risk cases.