OBJECTIVE: To describe the implementation and evaluation of a web-based medication error reporting system. DESIGN: Evaluation study. SETTING: Long-term care. PARTICIPANTS: 25 nursing homes in the US state of North Carolina. INTERVENTION: Detailed information about all medication errors occurring in a facility during a 1 year period was entered into a web-based reporting system. An evaluation survey was conducted to assess usability and the potential for the system to prevent errors. MAIN OUTCOME MEASURES: Number and specific characteristics of medication errors reported. A survey evaluating ease of use of the system and whether the participants thought it would help improve medication safety. RESULTS: 23 (92%) sites entered 631 error reports for 2731 discrete error instances when weighted by the number of times the errors were repeated. 51 (8%) errors were classified as having a serious patient impact requiring monitoring/intervention or worse. The most common errors were dose omission (203, 32%), overdose (91, 14%), underdose (43, 7%), wrong patient (38, 6%), wrong product (38, 6%), and wrong strength (38, 6%). Errors most commonly occurred during medication administration (296, 47%) and were attributed to basic human error (402, 48%). Seven drugs were implicated in a third (175, 28%) of all errors: lorazepam, oxycodone, warfarin, furosemide, hydrocodone, insulin and fentanyl. 20 sites (86% of respondents) completed the evaluation survey and participants found the system easy to use and thought it would increase accuracy of reporting and improve patient safety. CONCLUSIONS: The web-based medication error reporting system was easy to use, with strong indications that it would be a valuable tool for preventing future errors.