Fluoxetine (F) is a specific serotonin-reuptake inhibitor that has been shown to promote weight loss and improve glycemic control in obese diabetic patients. To study its long-term metabolic effect, 40 obese patients with non-insulin-dependent diabetes mellitus (NIDDM) or impaired glucose tolerance (IGT) were included in a 12-month, randomized, placebo-controlled study. Patients were assigned to receive either 60 mg F or placebo (P) daily in conjunction with a 5.0-MJ/d diet (> 50% carbohydrate). Both groups showed a significant weight loss, with a nadir after 6 months without group differences (mean ± SD: F, 10.1 ± 10.0 kg; P, 9.4 ± 11.5 kg). Fifteen patients from the F group and 14 from the P group completed the 12-month study without weight loss differences. Glycemic regulation improved along with the weight loss, but with a larger decline in plasma C-peptide and fasting glucose levels in the F group (P < .05). Total skeletal muscle glycogen synthase (GS) activity increased by 31% in the F group (P < .01) and by 17% in the P group (nonsignificant) after 6 months of treatment, but was still less than the activity in normal-weight controls (aged 28.0 ± 6.3 years; body mass index, 23.5 ± 2.2). After adjustment for fasting glucose, insulin, weight loss, and diabetic state, a positive effect of F remained on the total GS activity, which accounted for 27% of the variation (P < .05). The waist to hip ratio was reduced in P subjects as compared with F subjects (P < .05). Fat-free mass (FFM) tended to be more reduced in the F group as compared with P subjects (4.9 v 1.9 kg), although the difference did not reach statistical significance. In conclusion, F seems to improve insulin sensitivity beyond the effect mediated through weight loss by a possible effect on GS activity in skeletal muscle tissue.