To examine whether a physical activity program could improve physical fitness and glycemic control, 32 children and adolescents with insulin-dependent diabetes mellitus (IDDM) were examined before the program and 3 mo later. Fifty percent of the subjects (n = 16) participated in the training for 1 h/wk (exercise group), whereas the remaining subjects were engaged in nonphysical activities for an equal amount of time (nonexercise group). Age of the subjects ranged from 8.2 to 16.9 yr, (mean 11.9 yr), with mean duration of diabetes 0.6-13.1 yr (5.2 yr). During the 3-mo program peak oxygen consumption (VO2) rose from 40.0 to 43.8 ml.min-1.m-2 (P less than .01) in the exercise group but only by 1.3 ml.min-1.m-2 in the nonexercise group (NS). Metabolic control did not improve in either group, with glycosylated hemoglobin level rising from 9.8 to 10.5% (P less than .01) in the exercise group and from 9.4 to 9.7% (NS) in the control group. When subjects were stratified according to their participation, metabolic control was significantly better among diabetic subjects participating frequently (greater than or equal to 11 of 13 sessions) than among those participating infrequently (less than 11 of 13 sessions), regardless of the type of activity. It was concluded that a training program of 1 h/wk for 3 mo does improve physical fitness but not the metabolic control of diabetes. On the other hand, glycemic control appears to be best among diabetic subjects who are motivated to participate in any kind of program related to the treatment of their disease.