We compared pharmacokinetic parameters derived from three aminoglycoside serum concentration sampling methods and evaluated their effects on recommended aminoglycoside dosing regimens in 60 critically ill surgery patients. Patients had presumed or documented gram-negative sepsis, and had at least 4 aminoglycoside serum concentrations measured. We used a one-compartment model for peak and trough, 3-point series, and 4-point series sampling methods. Dosing regimens were calculated for each patient based on values derived from each method. We found differences in regimens for nearly 50% of patients if either 4- or 3-point series sampling was used to calculate the recommended dosage rather than peak and trough sampling. However, the 3-point method required a clinically significant change in regimen in only 12% of patients compared with 4-point sampling. The variability of all values derived from 3-point sampling were well accounted for by the 4-point method (r2 > 0.80). In addition, we noted significantly greater relative precision for 3-point sampling than peak and trough sampling for estimates of clearance, elimination rate, recommended daily dosage, and recommended dosing frequency. We recommend three optimally timed samples be drawn instead of peak and trough levels in dosing aminoglycosides in critically ill surgery patients.