At a health examination of men approximately 60 years old, 48 were found to have Type 2 (non-insulin-dependent) diabetes. All had been healthy 10 years earlier. The intention was to randomly allocate those with mild diabetes to a training or a control group and study the long-term influence of regular training on the progress of the disease. However, it was found that the majority--39 of 48--had other diseases or were on treatment which made regular training difficult and would have complicated the interpretation of the metabolic effects of training. Eight men took part in a physical training programme for up to 2 years. Aerobic capacity, glucose tolerance and serum lipoproteins were measured before and at the end of the training period. Another group of eight men with Type 2 diabetes, of similar weight, age and relative aerobic capacity, who did not undergo the physical training, served as control subjects. The oxygen uptake increased by 16% in the training group and decreased by 12% in the control group but no significant changes could be demonstrated in body weight, fasting blood glucose, serum lipoproteins, glucose tolerance or insulin values at a 75-g oral glucose tolerance test between the two groups. Two patients in the training group developed coronary heart disease, two deteriorated metabolically (drug treatment had to be added), and there were two drop-outs. The results cast doubts on the feasibility and efficacy of physical training as a long-term treatment for the majority of Type 2 diabetic patients, who are older than 60 years.