Ibuprofen was evaluated as an antipyretic agent in 178 children (aged 3 months to 12 years) to compare dosage (5 vs 10 mg/kg), establish absolute efficacy (with a placebo control group), determine relative efficacy (ibuprofen vs acetaminophen), evaluate maximum efficacy, and identify potential confounding variables. Ibuprofen 5 mg/kg was minimally effective in children less than 6 years of age who had an initial temperature of at least 38.8 degrees C (101.9 degrees F). Ibuprofen 10 mg/kg was more effective for febrile children. The area under the curve for temperature (or change in temperature) captured the net effect of each drug and provided the best estimate for efficacy comparison during a defined period. A linear correlation between initial temperature and measures of efficacy was observed. A twofold increase in efficacy was observed for children with an initial temperature less than 38.8 degrees C. A similar effect was noted for each treatment group. Age was also found to have confounding effects on antipyretic response. A complex interaction between antipyretic response, initial temperature, and age raises questions about the pharmacodynamics of the antipyretic response. We conclude that the most important variable in antipyretic study design is initial temperature. The influence of initial temperature on the magnitude of the response to an antipyretic drug is a previously unappreciated finding with potential impact on pharmacodynamic investigations of antipyretic medications. We describe this finding as nonlinear pharmacodynamics.