To examine the cellular mechanism of the antihyperglycemic action of in vivo metformin (M) we used an animal model of severe insulin resistance, the genetically obese (fa/fa) Zucker rat. The animals were treated with or without M (250 mg/kg.day) which was supplied with the drinking water. Three weeks of in vivo M-treatment had no effect on body weight and several blood lipid parameters, but markedly reduced plasma insulin levels by 45% (-M: 2932 +/- 166 vs. +M: 1614 +/- 85 pmol/liter, P < 0.01); plasma glucose was slightly but significantly decreased by 8.3% (-M: 7.2 +/- 0.2 vs. +M: 6.6 +/- 0.16 mmol/liter, P < 0.05). Adipocytes were isolated and incubated with or without insulin. In vivo M-treatment had no effect on basal 3-O-methylglucose uptake. In contrast, in vivo M-treatment increased insulin-stimulated glucose transport by 2.6 +/- 0.6-fold (P < 0.01). Measurement of cell surface insulin receptors revealed no effect of M on neither specific [125I]insulin binding nor on insulin receptor kinase activity. Insulin-mediated translocation of both GLUT1 and GLUT4 glucose transporters was enhanced by in vivo M-treatment, GLUT1 by 26.1%, GLUT4 by 30.5%. To fully account for the M-induced increment of insulin-stimulated glucose transport (2.6-fold), these data suggest that M increased the functional activity of glucose transporters. We conclude that amelioration of insulin resistance in (fa/fa) Zucker rats after 3 weeks of in vivo M-treatment is associated with 1) a marked reduction of in vivo hyperinsulinemia, 2) an increase of insulin-stimulated glucose transport in adipocytes; 3) this increase of insulin-stimulated glucose transport is accompanied with both a potentiation of insulin-induced translocation of GLUT1 and GLUT4 glucose transporters from an intracellular pool to the plasma membrane as well as increased functional activity of plasma membrane glucose transporters. 4) This M-effect seems to be independent of de novo glucose transporter synthesis, since total cellular GLUT1 and GLUT4 glucose transporter number were uneffected by M. 5) These results strongly suggest a direct action of M at the level of glucose transport, since neither tracer insulin binding nor insulin receptor kinase activity were significantly altered by M.