Metformin is a biguanide that has been shown to effectively lower plasma glucose levels in subjects with noninsulin-dependent diabetes mellitus (NIDDM). However, its mechanism of action remains unknown. Studies that have examined the effect of metformin on hepatic glucose production (HGP) and muscle glucose utilization in NIDDM have yielded conflicting results, and little information is available about the action of metformin on lactate turnover and gluconeogenesis from lactate in humans. We studied 20 NIDDM subjects and 8 nondiabetic controls in a randomized, double blind, placebo-controlled trial to determine the effect of 15 weeks of treatment with metformin or placebo on glucose and lactate metabolism. Before and after treatment, all participants received a 7-h infusion of [6-3H]glucose and [3-14C]lactate in combination with indirect calorimetry and estimation of lactate central vein specific activity. A euglycemic insulin clamp (20 mU/m2.min) was performed during the last 3 h of the tracer infusions. The study design allowed us to evaluate the effects of metformin vs. placebo treatment on glycemic control, plasma lipid profile, HGP, insulin-mediated glucose uptake, oxidative and nonoxidative glucose metabolism, and lactate turnover. Metformin treatment significantly reduced fasting plasma glucose (196 +/- 18 vs. 152 +/- 12 mg/dL; P < 0.01), hemoglobin A1 (12.5 +/- 0.6 vs. 9.2 +/- 0.3%; P < 0.01), and plasma triglyceride and low density lipoprotein cholesterol concentrations. When diabetics were compared to nondiabetic controls, basal HGP was higher (12.9 +/- 1.0 vs. 9.8 +/- 1.2 mumol/kg.min; P < 0.01) despite the presence of fasting hyperinsulinemia and insulin-mediated total body glucose disposal (10.9 +/- 0.9 vs. 20.2 +/- 3.3 mumol/kg.min; P < 0.01) was decreased. Metformin significantly reduced fasting HGP (from 12.9 +/- 0.7 to 11.0 +/- 0.5 mumol/kg.min; P < 0.01), but did not enhance total body glucose disposal during insulin stimulation (10.9 +/- 0.9 vs. 11.0 +/- 0.5 mumol/kg.min; P = NS). Neither oxidative nor nonoxidative glucose disposal was improved by metformin treatment. The fasting plasma lactate concentration (1.1 +/- 0.1 vs. 0.6 +/- 0.1 mmol/L) and lactate turnover (14.0 +/- 0.8 vs. 10.3 +/- 0.6 mumol/kg.min) were significantly increased in diabetics and strongly correlated (r = 0.68; P < 0.001). The percent gluconeogenesis derived from lactate was similar in diabetic and control subjects (17 +/- 2% vs. 15 +/- 2%; P = NS), but the estimated rate of gluconeogenesis from lactate was increased in the diabetic group (P < 0.01). Despite the significant reduction in HGP after metformin treatment, the percentage of gluconeogenesis from lactate and the rate of lactate-derived gluconeogenesis were unchanged from baseline. Basal lactate turnover (15.4 +/- 1.4 vs. 14.8 +/- 1.4 mumol/kg.min) and lactate oxidation (7.9 +/- 0.7 vs. 8.1 +/- 0.9 mumol/ kg.min) as well as total lactate turnover and lactate oxidation during the insulin clamp were similar before and after metformin treatment. There were no changes in any of the above metabolic parameters in the placebo-treated group. In poorly controlled NIDDM subjects, the primary mechanism by which metformin improves glycemic control is related to the suppression of accelerated basal HGP, and this most likely is secondary to an inhibition of hepatic glycogenolysis. Metformin has no effect on the rate of lactate turnover or gluconeogenesis from lactate in either the basal or insulin-stimulated states.