The global health care community has worked tirelessly for nearly a decade to make medical care safer for patients, but it still has limited ability to evaluate whether safety has improved. While there is a universal push to measure safety outcomes, the main barrier has been poor investment in the basic science of patient safety. This science would allow us to comprehend the causes of harm, design and pilot test interventions to reduce harm, and robustly evaluate their impact. This article describes several dilemmas in measuring patient safety, outlines a conceptual model and presents a framework for measuring patient safety, and offers future directions. Future research should seek to create a scientifically sound and feasible safety scorecard and improve performance.