Aminoglycosides are widely used, and clinicians continue to seek newer and better methods for initial dosing of these agents. Recently, three new methods were introduced: Thomson, Reesor Nimmo, and dosing in renopathy by easy-to-use multipliers (DREM). In comparing them with older, traditional dosing methods in patients with various degrees of renal function, the pharmacokinetic variables of gentamicin were determined from steady-state peak (Cmax) and trough (Cmin) serum concentrations using individualized regimens in 88 patients. Dosages were determined in each patient using the method of Hull-Sarubbi, rule of eights, and the three new methods, and the resultant Cmax and Cmin values were calculated from dosages generated by each method. The daily doses and Cmax values derived with the Hull-Sarubbi, Thomson, and Reesor Nimmo methods were not significantly different (p > 0.05). The Hull-Sarubbi was the most precise (root mean squared prediction error 1.3) and least biased (mean prediction error -0.05) of the five methods in predicting target gentamicin serum peak concentrations (Cmax 6.5 mg/L). The Hull-Sarubbi (69%), Thomson (86%), and Reesor Nimmo (70%) methods yielded therapeutic Cmax (5-8 mg/L) in a significantly higher percentage of patients than did the rule of eights (32%) and DREM (35%), (p < 0.05). Therefore, if gentamicin serum concentrations are not available, the first three appear to be reasonable methods for initiating gentamicin dosage regimens, but the last two may not be desirable to use in a clinical setting. These conclusions are based on the assumption that patients are adults with stable renal function and relatively stable clinical conditions.