The objective of this study was to explore racial differences in mortality for congenital heart surgery. We performed a population-based retrospective cohort study using hospital discharge abstract data from four states in 1996. The outcome measure was risk-adjusted in-hospital mortality. Cases of pediatric congenital heart surgery were classified into six risk categories using the Risk Adjustment in Congenital Heart Surgery method. Differences in risk-adjusted in-hospital mortality among racial groups were explored. Analyses stratified by state were used to identify regional differences. Of 5791 cases, 4822 (83%) were assigned to a risk group for analysis. Surgical mortality differed for whites compared to non-whites (3.7 vs 5.1%, p = 0.02). Among non-white groups, unadjusted mortality rates varied: Asian, 5.3%; black, 4.1%; Hispanic 4.9%; other, 7.3%; and missing, 7.6% (p = 0.008). Adjusted mortality also differed by race but was inconsistent across regions, making explanatory factors based solely on biology implausible. For example, compared to whites, blacks had a higher risk of dying in Massachusetts [odds ratio (OR) = 6.39, p = 0.08] but lower in Pennsylvania (OR = 0.41, p = 0.009). Adding insurance type to models did not eliminate racial differences. In risk-adjusted analyses, non-white groups had a higher risk of dying after congenital heart surgery than whites. Inconsistent effects among regions suggest that differential mortality is due to unequal access to care rather than biology.