The non-obese diabetic (NOD) mouse was used to investigate the effects of voluntary wheel running exercise on blood glucose levels, glycosylated hemoglobin, and longevity in Type 1 diabetes mellitus. In Experiment 1, diabetic and normoglycemic mice exercised 5 h/day, 5 days/week for 3 weeks matched with non-exercising controls. In diabetic animals a positive correlation was found between blood glucose and the number of revolutions performed (P < or = 0.02). Exercise also significantly lowered blood glucose between baseline and post-exercise in both diabetic and normoglycemic animals. In Experiment 2, mice exercised 2 h/day, 5 days/week. For the diabetic animals, glycosylated hemoglobin was lower than that of matched non-exercising diabetic animals at week 3 (11.1 +/- 0.6% vs. 15.0 +/- 1.6%, P < 0.001). Diabetic runners were able to train and significantly increased running in the first 4 weeks (P < 0.05). At the end of 9 weeks all 5 diabetic runners were alive, compared with 3 of 5 non-running diabetic animals. We conclude: (i) the NOD mouse is a useful model for the study of exercise in Type 1 diabetes, (ii) running exercise is associated with a drop in blood glucose, (iii) the amount of voluntary exercise performed correlates with blood glucose in diabetic animals, and (iv) diabetic mice will increase running distance in the first few weeks after diagnosis.