BACKGROUND: Hospitals show wide variation in outcomes and systems of care. It is unclear whether hospital complexity-the range of services and technologies provided-affects outcomes and in what direction. We sought to determine whether complexity was associated with inpatient surgical mortality. METHODS: Using national Medicare data, we identified all fee-for-service inpatients who underwent 1 of 5 common high-risk surgical procedures in 2008-2009 and measured complexity by the number of unique primary diagnoses admitted to each hospital over the 2-year period. We calculated 30-day postoperative mortality rates, adjusting for patient and hospital characteristics, and used multivariable Poisson regression models to test for an association between hospital complexity and mortality rates. We then used this model to generate predicted mortality rates for low-volume and high-volume hospitals across the spectrum of hospital complexity. RESULTS: A total of 2691 hospitals were analyzed, representing a total of 382,372 admissions. After adjusting for hospital characteristics, including hospital volume, increasing hospital complexity was associated with lower surgical mortality rates. Patients receiving care at the hospitals in the lowest quintile of unique diagnoses had a 27% higher risk of death than those at the highest quintile. The effect of complexity was largest for low-volume hospitals, which were capable of achieving mortality rates similar to high-volume hospitals when in the most complex quintile. CONCLUSIONS: Hospital complexity matters and is associated with lower surgical mortality rates, independent of hospital volume. The effect of complexity on outcomes for nonsurgical services warrants investigation.