Health care institutions are still searching for the best approaches to prospectively identify safety hazards before patients are harmed and, when errors do occur, the optimal methods to recognize them as rapidly as possible to prevent further harm. This commentary elegantly makes the point that individual methods of identifying safety incidents provide only a limited perspective on overall institutional safety, and therefore multiple methods of amassing and analyzing safety data are necessary in order to provide a comprehensive picture of safety issues. The article discusses the merits and limitations of 18 different data sources, each of which can yield useful information on specific types of harm, but none of which should be used as the sole source of safety data.