Medication errors are preventable adverse events or unsafe conditions caused by inappropriate uses of medication. To collect data of patient safety events (PSE) and to analyze the root causes of PSE, reporting systems have been implemented in healthcare settings and patient safety organizations (PSO). However, the poor data quality of reports impedes the reporting and root cause analysis (RCA) of PSE. Incomplete or missing data is the most prevalent problem in event reports. To assess the data quality of PSE reports, we used an adapted taxonomy as the data evaluation model to evaluate the quality of narrative reports collected by a PSO. Sample reports were extracted based on eight error types and scored by experts. Most structured fields in the reports were ignored by reporters. In contrast, the narrative parts of the reports contain rich and valuable information. The evaluation results show that the adapted taxonomy is a promising tool for report quality assessment and improvement.