The development of accurate and reliable measurements was identified as a major priority for the patient safety field in an influential 2015 report. This mixed-methods study of a large-scale improvement program in the United Kingdom provides important insights into the challenges of measuring safety in real-world settings. In the program, quality improvement teams at each of the nine participating hospitals chose targets and developed metrics with assistance from external mentors. The measurement strategies were reviewed by the study investigators, who also conducted structured interviews with quality improvement team members at each site. Measurement was a challenge for all sites, attributed in large part to frontline staff's lack of experience in data analysis and reliance on homegrown rather than externally validated safety metrics. This manifested in the use of metrics that often were overambitious, not linked to the interventions that were being implemented, or not analyzed in a statistically appropriate fashion. As a result, it was difficult to determine if improvement in the safety targets had been achieved. A previous PSNet interview featured the study's senior author, Mary Dixon-Woods.