The effects of renal impairment and age on the pharmacokinetics of metformin were evaluated. The subjects, including 6 young, 12 elderly, and 3 middle-age healthy adults and 15 adults with various degrees of chronic renal impairment (CRI) each were given a single, 850-mg metformin HCl tablet. Multiple whole blood, plasma, and urine samples were collected and analyzed for metformin levels using a high-performance liquid chromatography (HPLC) method. In healthy elderly individuals, the plasma and whole blood clearance/absolute bioavailability values [CL/F and (CL/F)b], and corresponding renal clearance values (CLR and CLR,b) of metformin were 35-40% lower than the respective values in healthy young individuals. These two groups did not differ significantly with respect to volume of distribution (Vd), time to peak concentration (tmax), and parameters related to metformin's appearance in the urine. In the moderate and severe CRI groups, all clearance values were 74-78% lower than in the healthy young/middle-age group, and all other evaluable pharmacokinetic parameters (with the exception of tmax) differed significantly in this group. In the mild CRI group, clearance values of metformin, which were 23-33% lower than in the young/middle-age group, were the only parameters that differed significantly. Based on a regression analysis of the combined data, both creatinine clearance (CL*cr; corrected for body surface area) and age are predictors of metformin clearance, with the following model best fitting the data: CL/F [or (CL/F)b, CLR, CLR,b] = alpha + beta.CL*cr + gamma.CL*cr.age. Metformin should not be used in patients with moderate and severe CRI, or in patients with mild, but not absolutely stable, renal impairment. The initial and maximum doses in elderly patients and patients with stable mild CRI should be lowered to approximately one third that given to the general (i.e., patients without non-insulin-dependent diabetes) population.