We estimate doctor value-added and provide evidence on the distribution of physician quality in an entire country, combining rich population-wide register data with random assignment of patients to general practitioners (GPs). We show that there is substantial variation in the quality of physicians, as measured by patients’ post-assignment mortality, in the primary care sector. Specifically, a one standard deviation increase in doctor quality is associated with a 12.2-percentage point decline in a patient’s two-year mortality risk. While we find evidence of observable doctor characteristics and practice styles influencing a GP’s value-added, a standard decomposition exercise reveals that most of the quality variation is driven by unobserved differences across doctors. Finally, we show that patients are unable to identify who the high-quality doctors are, and that patient-generated GP ratings are uncorrelated with GP value-added. Using a lower bound of the predicted value of an additional life year in Norway ($35,000), our results demonstrate that replacing the worst performing GPs (bottom 5 percent of the VA distribution) with GPs of average quality generates a social benefit of $27,417 per patient, $9.05 million per GP, or $934 million in total. At the same time, our results show that higher-quality GPs are associated with a lower per-patient cost.