Using a rich sample of admission records from New Orleans Touro Infirmary, we examine the in-hospital mortality risk of free and enslaved patients. Despite a higher mortality rate in the general population, slaves were significantly less likely to die in the hospital than the whites. We analyze the determinants of in-hospital mortality at Touro using Oaxaca-type decomposition to aggregate our regression results. After controlling for differences in characteristics and maladies, we find that much of the mortality gap remains unexplained. In conclusion, we propose an alternative explanation for the mortality gap based on the selective hospital admission of slaves.