We retrospectively analyzed amikacin pharmacokinetics in 28 patients (mean age, 47.4 +/- 13.6 years) who received high-dose chemotherapy during a neutropenic febrile episode. Patients received an experimental protocol of high-dose anticancer chemotherapy. Amikacin pharmacokinetic parameters were calculated from two or more concentrations in serum around a single dose by the method of Sawchuck and Zaske (J. Pharmacokinet. Biopharm. 4:183-195, 1976). Predicted parameters were calculated by using standard methods. The observed amikacin volume of distribution and clearance were significantly greater and the elimination half-life was longer than predicted (0.38 +/- 0.13 versus 0.25 liter/kg [P = 0.0001], 1.51 +/- 0.92 versus 1.17 +/- 0.38 liters/h/kg [P = 0.012], and 3.8 +/- 2.4 versus 2.9 +/- 1.1 h [P = 0.011], respectively). Multivariate analysis revealed that albumin correlated negatively and creatinine correlated positively with the volume of distribution and the elimination half-life. Creatinine and the percentage below the ideal body weight correlated negatively and hematocrit correlated positively with clearance. Administration of dosage regimens based on predicted pharmacokinetic parameters yielded subtherapeutic amikacin concentrations in serum in our patients. Because of the increased dosage requirements and the need for adequate antibiotic treatment in this population, we suggest guidelines for empiric dosing for patients with advanced cancer receiving intensive chemotherapy.