Intrapatient variation in aminoglycoside pharmacokinetics was studied in 100 critically ill surgery patients, and the relationships between aminoglycoside pharmacokinetics and physiological variables were evaluated. A one-compartment model and the DataMed Clinical Support Service System were used to calculate pharmacokinetic values from serum concentration-time data and to provide individualized dosage regimens. A total of 323 pharmacokinetic analyses (2 to 9 per patient) were performed. There were significant intrapatient differences in clearance (CL) between analyses, and there was a downward trend in elimination rate constant (k) and an upward trend in half-life (t1/2) with time. There were no significant differences in mean volume of distribution (V) between analyses and no strong correlations of V, k, and CL between analyses. The total daily dosage necessary to provide the desired serum aminoglycoside concentration decreased with time, while the dosing interval increased. The mean absolute changes in t1/2, V, and CL between analyses were 3.2 hr, 7.1 L, and 15.7 mL/min, respectively. Dosing regimen changes were required in 90% of the patients, regardless of the time between analyses, including 75% (24/32) of the patients with analyses on consecutive days. There were weak correlations between changes in serum creatinine concentration and changes in dosing interval and between changes in weight and changes in V. Changes in calculated creatinine clearance, BUN, pulmonary capillary wedge pressure, central venous pressure, systemic vascular resistance, cardiac index, input minus output, and maximum temperature did not result in strong correlations with k, t1/2, V, and CL by univariate or multivariate regression analyses.(ABSTRACT TRUNCATED AT 250 WORDS)