Measurement of patient safety serves to identify opportunities to improve safety within a neonatal intensive care unit (NICU), compare the safety of care provided by different NICUs, determine changes in response to safety interventions or programs, follow safety trends over time, and potentially deny payment for specific events. The ideal patient safety measures are rates of events derived from surveillance with valid and reliable detection of numerators (errors or adverse events) and denominators (the opportunities for errors or adverse events to occur). Methods used to identify these numerators and denominators include reporting, direct observation, videotaping, chart review, trigger tools, and automated methods. However, there are significant methodological and practical (feasibility) challenges to the accurate and reliable determination of rates of errors and adverse events. These include failure to detect and document such events, surveillance bias, lack of consistent definitions, frequent requirement for judgment in identifying and classifying challenges (which introduces interrater inconsistency), and need for significant additional resources.