The present studies have been designed to evaluate the effects of physical training in rats with a diminished insulin reserve. Mild diabetes mellitus was induced in rats with 45 mg/kg streptozotocin. Physical training was done on a treadmill, with a progressive program, twice daily, 5 days per week, for 10 wk in control and diabetic rats. At the end of the training program, a significant diminution in body weight gain and in epididymal fat pad weight was observed in both trained groups, as compared with sedentary controls. Sixty-four hours after the last exercise, control (N = 16), control-trained (N = 14), diabetic (N = 17), and diabetic-trained (N = 15) rats were submitted to an intravenous glucose tolerance test (0.5 g/kg). Arterial blood samples were collected at -15, 0, 2, 4, 6, 10, 15, 30, 45, and 60 min during the test in unanesthetized and precannulated rats for plasma glucose and insulin determinations. In normal rats, physical training induced a sharp decrease in the basal insulin levels (36 +/- vs. 101 +/- 6 microunits/ml; P less than 0.001) without any significant changes in glucose levels (122 +/- 4 vs. 129 +/- 2 mg/dl; P less than 0.05). After the glucose loading there was no significant change in the glucose tolerance curve, although the insulin values remained lower throughout the test in the trained group. In the diabetic rats, the elevated basal glucose levels were significantly diminished in the trained group as compared with the untrained diabetic group (177 +/- 22 vs. 306 +/- 37 mg/dl; P less than 0.001), although the basal insulin values were similar in both groups (51 +/- 7 vs. 54 +/- 9 microunits/ml; P greater than 0.05). The improvement in the glucose tolerance of the diabetic-trained rats was further confirmed by the glucose disappearance rate constant that was significantly increased (3.6 0.4 vs. 2.0 +/- 0.3; P less than 0.01), although not fully restored to normal (6.3 +/- 0.2; P less than 0.001). These data clearly show that in rats with a diminished insulin reserve, a 10-wk running program greatly improved the glucose homeostasis. Measurements of circulating insulin suggest that, although an effect on insulin secretion cannot be totally excluded, the beneficial effect of physical training on diabetes mellitus is probably best explained by an increase in insulin sensitivity.