Metformin (Met) is a biguanide oral hypoglycemic agent used in the treatment of noninsulin-dependent diabetes mellitus (NIDDM). To define whether the glucose-lowering effects are mediated via alterations of insulin receptors, the effects of Met in vitro in rat adipocytes and in vivo in patients with poorly controlled NIDDM were studied. In vitro exposure of rat adipose tissue to metformin for 20 h resulted in a significant increase in insulin binding (mean +/- SEM percent specific [125I]insulin bound per 10(5) adipocytes: control, 1.35 +/- 0.13; Met, 1.69 +/- 0.18; P less than 0.02). No change occurred after 2 h of exposure or less. In contrast, after only 1 h of preincubation. Met alone stimulated [U-14C]glucose oxidation by 58 +/- 15.5% (P less than 0.01). Met did not stimulate glucose oxidation in the presence of a high insulin concentration. For the in vivo studies, oral glucose tolerance tests and monocyte [125I]insulin binding assays were performed before and after 7 days of Met treatment (2 g/day) in 18 patients with poorly controlled NIDDM. All patients responded to Met with a decrease in fasting and postglucose plasma glucose concentrations, but no change in insulin concentrations [pre-Met vs. post-Met: fasting plasma glucose, 210 +/- 10 vs. 157 +/- 11 mg/dl (P less than 0.001); fasting plasma insulin, 20.3 +/- 3.1 vs. 18.4 +/- 2.0 microU/ml]. When insulin binding was examined, 8 patients with decreased binding each responded to Met with a 50% or greater increase (group 1), while 10 patients with normal binding had no increase after treatment (group 2). However, both groups had similar lowering of glucose concentrations [fasting plasma glucose: group 1, 205 +/- 19 vs. 153 +/- 20 (P less than 0.001); group 2, 214 +/- 11 vs. 160 +/- 13 (P less than 0.001)]. We conclude that 1) Met has an acute insulin-like effect in vitro independent of its ability to increase insulin binding; 2) Met acts in vivo predominantly at a postreceptor site to lower plasma glucose; 3) the glucose-lowering effect is independent of pretreatment insulin binding status; and 4) the increase in insulin binding after Met treatment in patients with NIDDM and low insulin binding occurs without changes in insulin concentrations.