Insulin resistance contributes to the metabolic defects in non-insulin dependent diabetes mellitus (NIDDM). Anorectic agents have been shown to improve insulin action in NIDDM, irrespective of weight reduction. In a double-blind placebo-controlled cross-over study, we examined hepatic and peripheral insulin action by the sequential hyperinsulinaemic-euglycaemic clamp technique with infusion of 3-[3H]-glucose in eight obese NIDDM patients and in eight obese non-diabetics, matched for age, sex and body mass index. Body weight was kept constant. After 14 days of fluoxetine, 60 mg daily, in NIDDM half-maximal peripheral glucose uptake was achieved at a lower insulin level than after placebo (ED50pgu: 180.5 +/- 25.8 vs. 225.3 +/- 39.9 mU/l, P < 0.05), but not in non-diabetics (140 +/- 15.3 vs. 135.3 +/- 22.2 mU/l, n.s.). Maximal peripheral glucose uptake (Vmaxpgu) did not change significantly. Multivariate analysis disclosed no differences in the effect of fluoxetine between NIDDM and non-diabetics. When non-diabetics and NIDDM were considered together, only the most insulin-resistant individuals demonstrated a decrease in ED50pgu (P < 0.001). Likewise, only the individuals with the most outspoken hepatic insulin resistance demonstrated a decrease in insulin level, at which hepatic glucose production (HGP) is completely suppressed (HGP0) (P < 0.01). In conclusion, fluoxetine improves peripheral and hepatic insulin action in obese insulin-resistant subjects irrespective of its weight lowering effect.