This study sought to establish whether the glycemic control achieved in exercise-trained adolescents with type 1 diabetes mellitus (DM) is dependent on the quality of glycemic control prior to the initiation of exercise training. Adolescents with type 1 DM were randomly assigned to groups with either lower or higher than 9% glycosylated hemoglobin (HbA1c), and submitted to 12 weeks of supervised training followed by 12 weeks of unsupervised training (n = 12 per group). Supervised training caused a 17% rise in the patients' aerobic capacity which during the ensuing period of unsupervised training decreased to pre-training levels, thus suggesting a poor compliance with unsupervised training. The average levels of HbA1c in poorly and well controlled diabetic patients were not affected by training, a finding indicating that irrespective of the quality of glycemia prior to exercise training, glycemic control in adolescents with type 1 DM does not improve in response to exercise training alone.