To determine the effects of weight loss on insulin action in patients with non-insulin-dependent diabetes mellitus (NIDDM) and in their isolated adipocytes, we studied nine weight-stabilized Pima Indians [7 females and 2 males; age 39 +/- 3 yr; wt 99.9 +/- 8.2 kg; body fat 39 +/- 2% (means +/- SE)] before and after a 6.7 +/- 1.3-kg weight loss and decrease in fasting plasma glucose from 250 +/- 11 to 148 +/- 15 mg/dl. In vivo insulin action was measured during a 3-insulin-step, hyperglycemic (approximately 310 mg/dl) clamp with somatostatin (250 micrograms/h). At a clamp plasma insulin concentration of 10 microU/ml, glucose disposal rates did not change after weight loss; at approximately 100 microU/ml, glucose disposal rates increased by 21% [from 4.3 +/- 0.2 to 5.3 +/- 0.4 mg X min-1 X kg-1 of fat-free mass (FFM), P less than .01] mostly due to increased carbohydrate oxidation rates (2.0 +/- 0.3 to 2.8 +/- 0.3 mg X min-1 X kg-1 FFM, P less than .02); at 2400 microU/ml, glucose disposal rates increased by 37% (11.4 +/- 0.6 to 15.6 +/- 1.4 mg X min-1 X kg-1 FFM, P less than .02) mostly due to increased nonoxidative carbohydrate disposal rates or storage (7.5 +/- 0.6 to 10.9 +/- 1.3 mg X min-1 X kg-1 FFM, P less than .04). Sensitivity of glucose disposal to insulin in the physiologic range (measured as change in glucose disposal rate per unit change in insulin concentration between clamps at approximately 10 and approximately 100 microU/ml) was very low in these diabetic subjects and did not change after weight loss. Adipocyte cell size, basal and maximal insulin-stimulated glucose transport, and half-maximal rate for transport did not change after weight loss. The data suggest that insulin in the physiologic range has no apparent effect on glucose disposal in patients with NIDDM before or after weight loss. However, a moderate weight loss is associated with enhanced capacity to transport and metabolize glucose in vivo. The discrepancy between in vivo and in vitro results suggests that the adipocyte may not always reflect in vivo insulin action. Diabetes 36:227-36, 1987.