Pharmacokinetic principles were used to determine amikacin doses for granulocytopenic cancer patients on empirical antibiotic regimens, and audiometry was used to determine the effect of this treatment on auditory acuity. Patients received ticarcillin 300 mg/kg/day plus amikacin, or moxalactam 8 g/day plus amikacin, in a blinded study. Amikacin doses were calculated to achieve a peak serum concentration (one hour after infusion) of approximately 25 micrograms/ml and a trough concentration of approximately 8 micrograms/ml. Baseline audiometry was conducted, and follow-up audiometry was done on completion of the antibiotic therapy. A decrease of greater than or equal to 20 db at any frequency in one or both ears was considered indicative of ototoxicity. Of 201 patients on the empirical antibiotic protocol, 55 had courses of treatment that could be evaluated. Ototoxicity not attributable to any other drug or disease process occurred in 10 patients. There were no significant differences related to age, weight, daily dose, or total dose between patients who developed ototoxicity and those who did not. Peak and trough amikacin concentrations were not significantly different between groups. More women than men developed ototoxicity. Duration of therapy for the ototoxic group was longer, and there was a trend for those with abnormal initial audiograms to develop further evidence of impairment. The significant risk factor for auditory toxicity was the combination of duration of aminoglycoside therapy and total dose per kilogram. Pharmacokinetically calculated doses of amikacin did not result in a higher incidence of auditory toxicity than reported in previous studies.