Few data exist concerning the effect of obesity on the pharmacokinetic parameters of vancomycin. The purpose of this investigation was to assess the effect of obesity on vancomycin pharmacokinetic parameters in 95 nonobese and 135 obese adult patients (age range, 18 to 92 years) receiving vancomycin. All subjects had normal renal function as defined by a creatinine concentration in serum of ≤1.5 mg/dl (mean estimated creatinine clearance ± 1 standard deviation, 76 ± 34; range, 23 to 215 ml/min). Vancomycin concentrations in serum were determined by the fluorescence polarization immunoassay. All data for vancomycin concentration in serum versus time for each course of therapy were fitted by using a two-compartment Bayesian forecasting program. Subjects were stratified into nine groups on the basis of the percent difference between actual body weight (ABW) and lean body weight (LBW) (>-10%, -10 to 0%, >0 to 10%, >10 to 20%, >20 to 30%, >30 to 40%, >40 to 50%, >50 to 60%, >60%). Analysis of variance with post hoc Scheffe's testing revealed that statistically significant differences occurred in terminal disposition half-life (t( 1/2 β)) between the extremes of modestly obese (group 4) and morbidly obese (group 9, P < 0.05) patients. Similar analysis with distribution volume (V) identified significant differences in patients at or near their LBW (groups 2 to 4) and patients who were morbidly obese (groups 8 and 9, P < 0.05). Multiple regression models for the pharmacokinetic parameters V, t( 1/2 β), and vancomycin total body clearance were developed to assess the joint predictive power of LBW, ABW, and percent over LBW, controlling for the effects of age, initial creatinine concentration in serum, initial creatinine clearance, and gender. In the final model for V, both ABW and percent over LBW were independent and significant predictors. For total body clearance, only ABW was significant and predictive. Percent over LBW was a significant and independent predictor of t( 1/2 β). LBW is not predictive of these pharmacokinetic parameters and should not be used for initial dosing. On the basis of these data, ABW appears to be superior to LBW for calculating initial dose requirements for vancomycin.