The population pharmacokinetics of amikacin were studied in 40 geriatric medicine patients (aged 59-95 years, average 78 years) by the NPEM-2 algorithm, using a 2-compartment model. The fitted parameters were: renal clearance of amikacin, expressed as a fraction of creatinine clearance (CLS) and volume of the central compartment, expressed as a fraction of body weight (VS). The nonrenal clearance of amikacin (CLi) and the transfer constants between the 2 compartments (k12 and k21) were held constant. The distribution of each pharmacokinetic parameter was characterized by the median and the 50% dispersion factor (DF50) which is the interval between the 25th and 75th percentiles, divided by 1.32. The parameters were thus estimated by the following values: CLs = 0.91 +/- 0.45 and Vs = 0.29 +/- 0.10 l x kg-1. The volume of distribution was increased with respect to the usual value of 0.20 l x kg-1. The interindividual variability was high, particularly for clearance. Although the median value of CLs was close to unity, indicating that for most patients the elimination kinetics of amikacin followed that of creatinine, there was a slight probability to observe much higher values of CLs (up to 5), indicating that for some aged patients the elimination of amikacin was less altered than that of creatinine. These parameters were used as reference population values to estimate the pharmacokinetics of amikacin in a second group of 20 patients by the Bayesian method, with 2 blood samples per patient. For each patient the fitted parameters were able to predict the plasma concentrations of amikacin during the next 72 hours with no significant bias and a precision of 3.5 mg x 1(-1). This study confirms the ability of the NPEM-2 algorithm to provide reference population values for use in Bayesian monitoring of aminoglycoside therapy.