We report the use of a new method to determine patient population pharmacokinetic parameters (nonparametric expected maximum or NPEM). Our purpose was to develop and then analyze the utility of these parameters, compared to a more traditional approach. Nineteen patients with acute cholecystitis made up the control group for defining the parameters via NPEM. The standard of practice was to use a model created from a different intraabdominal infection group (appendicitis), referred to as "surgical patient model." These two models were compared with a group of 23 patients receiving gentamicin for acute cholecystitis. We concluded that the NPEM model was superior to the surgical patient model in predicting gentamicin trough and peak levels with less bias and better precision.