Four methods for calculating gentamicin sulfate dosage requirements were evaluated in 96 patients and compared with an individual method. The pharmacokinetic parameters of gentamicin were determined from serum concentration time data and used to calculate individualized dosage regimens. Doses were determined in each patients using the 'predictive methods' (Sarubbi-Hull, Dettli, 'rule of eights', and Chan). Resultant serum concentrations were calculated from doses arrived at by each method. These dosing methods resulted in a large proportion of patients with subtherapeutic or potentially toxic concentrations, or both. The Dettli and Chan methods produced therapeutic concentrations in more patients than the Sarubbi-Hull and rule of eight methods. Desired therapeutic concentrations were attained in significantly more patients with the individualized method than with the predictive methods, and, in addition, larger doses were required. The use of predictive dosage methods should be followed with serum concentration determinations and dosage adjustment to ensure therapeutic concentrations early in treatment.