Six men and three women with insulin-dependent diabetes (without complications) participated in physical training three times a week for 20 weeks. Physical training did not change the concentration of fasting blood-glucose, glucose excretion in urine or glucosylated haemoglobin (HbA1). However, the glucose disposal rate during euglycaemic clamp increased after training. In two patients a minor reduction of insulin dosage was necessary to alleviate slight hypoglycaemic episodes. The training resulted in significant increases in quadriceps isometric and dynamic strength and endurance. Maximal oxygen uptake increased by 8%, the activity of glycolytic enzymes in vastus lateralis muscle by 47% for hexokinase, and 30% for tri-osephosphate dehydrogenase and 25% for lactic dehydrogenase, the activity of oxidative enzymes by 42% for citrate synthase and 46% for 3-hydroxy-acyl-CoA-dehydrogenase. The glycogen concentration in the vastus lateralis muscle did not change significantly. Lipoprotein lipase activity did not change in muscle, nor in adipose tissue. The mean muscle fibre area increased by 25% and the area of FTa fibres by 30%. The new formation of capillaries around different muscle fibres was significant for FTb fibres (26%). The proliferation of capillaries, however, appeared to be insufficient to cope with the increased area of muscle fibres. As a result, the mean area of muscle fibre supplied by one capillary (a measure of diffusion distance) significantly increased after training for FTa fibres. It is concluded that with the exception of deficient proliferation of capillaries, patients with insulin-dependent diabetes mellitus show a normal central and peripheral adaptation to physical training. Physical training does not apparently improve blood glucose control in most cases, despite an increased insulin sensitivity.