To clarify the impact of vigorous physical training on in vivo insulin action and glucose metabolism independent of the intervening effects of concomitant changes in body weight and composition and residual effects of an acute exercise session, 10 lean, 10 obese, and 6 diet-controlled type II diabetic men trained for 12 wk on a cycle ergometer 4 h/wk at approximately 70% of maximal O2 uptake (VO2max) while body composition and weight were maintained by refeeding the energy expended in each training session. Before and 4-5 days after the last training session, euglycemic hyperinsulinemic (40 mU.m2.min-1) clamps were performed at a plasma glucose of 90 mg/dl, combined with indirect calorimetry. Total insulin-stimulated glucose disposal (M) was corrected for residual hepatic glucose output. Body weight, fat, and fat-free mass (FFM) did not change with training, but cardiorespiratory fitness increased by 27% in all groups. Before and after training, M was lower for the obese (5.33 +/- 0.39 mg.kg FFM-1.min-1 pretraining; 5.33 +/- 0.46 posttraining) than for the lean men (9.07 +/- 0.49 and 8.91 +/- 0.60 mg.kg FFM-1.min-1 for pretraining and posttraining, respectively) and lower for the diabetic (3.86 +/- 0.44 and 3.49 +/- 0.21) than for the obese men (P less than 0.001). Insulin sensitivity was not significantly altered by training in any group, but basal hepatic glucose production was reduced by 22% in the diabetic men. Thus, when intervening effects of the last exercise bout or body composition changes were controlled, exercise training per se leading to increased cardiorespiratory fitness had no independent impact on insulin action and did not improve the insulin resistance in obese or diabetic men.