The effects of 6 weeks of thrice weekly training on glycaemic control were assessed in 20 sedentary Type 2 (non-insulin-dependent) diabetic patients and 11 control subjects matched for previous physical activity. Maximal oxygen uptake was lower in the diabetic patients than in control subjects before training (26.2 +/- 1.1 versus 32.6 +-/ 1.7 ml X kg-1 X min-1; p less than 0.001). Glycosylated haemoglobin levels decreased in the diabetic patients during the training programme (12.2 +/- 0.5 to 10.7 +/- 0.4%; p less than 0.02). Oral and intravenous glucose tolerance determined 72 h after the last exercise period showed only minimal improvement. Plasma glucose levels were, however, significantly lower at 12 h than 72 h after exercise in eight subjects tested at both time points. These data suggest than an exercise programme can produce a significant decrease in glycosylated haemoglobin levels in Type 2 diabetic males probably due, in great measure, to the cumulative effect of transient improvements in glucose tolerance which follow each individual period of exercise.