Studies of the effect of computerized physician order entry (CPOE) in pediatrics have reached conflicting results. This study evaluated the impact of a commercial CPOE system on medication errors over a 16-month time frame. Strengths of this study included use of a comprehensive error-finding method and a time-series analytical approach. Implementation of the CPOE system was associated with a modest reduction in "nonintercepted" serious medication errors (errors that were not detected, and thus affected the patient), but there was no overall decrease in injuries suffered due to adverse drug events. Notably, the rate of dosing errors—the most common type of pediatric drug error—did not decrease, despite specific features of the CPOE system designed to prevent such errors. The authors discuss their experience with the system and subsequent modifications that have been made in response to these results.