This study was conducted to assess the effect of noninsulin-dependent diabetes mellitus (NIDDM) and gender on the pharmacokinetics of metformin and to investigate whether or not metformin exhibits dose-dependent pharmacokinetics. The pharmacodynamic effects (on plasma glucose and insulin) of metformin in patients with NIDDM and in healthy subjects also were assessed. Nine patients with NIDDM and 9 healthy subjects received 4 single-blind single-dose treatments of metformin HCL (850 mg, 1,700 mg, 2,550 mg, and placebo) and a multiple-dose treatment of 850 mg metformin HCL (3 times daily for 19 doses). After each single-dose treatment and the final dose of the multiple-dose phase, multiple plasma and urine samples were collected for 48 hours and assayed for metformin levels. Plasma samples were also assayed for glucose and insulin levels. There were no significant differences in metformin kinetics in patients with NIDDM compared with healthy subjects, in men compared with women, or during multiple-dose treatment versus single-dose treatment. Plasma concentrations of metformin increase less than proportionally to dose, most likely due to a decrease in percent absorbed. In patients with NIDDM, single doses of 1,700-mg or higher of metformin significantly decrease postprandial, but not preprandial, glucose concentrations and do not influence insulin concentrations. With multiple doses, both preprandial and postprandial glucose concentrations and preprandial insulin concentrations were significantly lower than with placebo. The effect of metformin on glucose level is correlated with the average fasting plasma glucose level without drug. In healthy subjects, single and multiple doses of metformin showed no effect on plasma glucose, but significantly attenuated the rise in immediate postprandial insulin levels.