The safety and efficacy of vanadyl sulfate (VS) was tested in a single-blind, placebo-controlled study. Eight patients (four men and four women) with non-insulin-dependent diabetes mellitus (NIDDM) received VS (50 mg twice daily orally) for 4 weeks. Six of these patients (four men and two women) continued in the study and were given a placebo for an additional 4 weeks. Euglycemic-hyperinsulinemic clamps were performed before and after the VS and placebo phases. VS was associated with gastrointestinal side effects in six of eight patients during the first week, but was well tolerated after that. VS administration was associated with a 20% decrease in fasting glucose concentration (from 9.3 +/- 1.8 to 7.4 +/- 1.4 mmol/L, P < .05) and a decrease in hepatic glucose output (HGO) during hyperinsulinemia (from 5.0 +/- 1.0 pre-VS to 3.1 +/- 0.9 micromol/kg x min post-VS, P < .02). The improvement in fasting plasma glucose and HGO that occurred during VS treatment was maintained during the placebo phase. VS had no significant effects on rates of total-body glucose uptake, glycogen synthesis, glycolysis, carbohydrate (CHO) oxidation, or lipolysis during euglycemic-hyperinsulinemic clamps. We conclude that VS at the dose used was well tolerated and resulted in modest reductions of fasting plasma glucose and hepatic insulin resistance. However, the safety of larger doses and use of vanadium salts for longer periods remains uncertain.