The objectives of these analyses were to (1) develop a population pharmacokinetic/pharmacodynamic model for a novel COX-2 inhibitor (CS-706) using data from primarily Caucasian subjects, (2) predict responses in subpopulations of interest (including Japanese subjects), and (3) correlate pharmacodynamic parameters to safety outcomes. The model was developed using data from 130 healthy adults following single or multiple doses of CS-706. Serial plasma concentrations of CS-706 and ex vivo whole-blood cyclooxygenase-1 (COX-1) and COX-2 activity were determined up to 72 hours postdose. An E(max) model described relationships between CS-706 plasma concentrations and COX-1 and COX-2 inhibition. CS-706 potency (EC(50)) was 397 ng/mL for COX-1 and 20 ng/mL for COX-2. None of the tested covariates influenced the pharmacodynamics of CS-706. Japanese subjects are expected to show a slightly reduced response to CS-706, consistent with lower exposure following the same dose given to Caucasian subjects. Predictive pharmacokinetic/pharmacodynamic modeling for COX-1 and COX-2 inhibition indicates a 20-fold potency ratio that is expected to be similar in Japanese and Caucasians. There was good correlation between COX-1 inhibition and the incidence of 7-day gastroduodenal mucosal injury. A dose of less than 25 mg bid could be adequate to inhibit COX-2 activity with a low risk of gastrointestinal mucosal injury.