We evaluated the association between severity of sepsis and in-hospital mortality in 150 patients with non-surgical sepsis at a regional referral hospital in Uganda. In-hospital mortality occurred in 5 of 52 (9.6%) patients with sepsis, 24 of 71 (33.8%) patients with severe sepsis, and 16 of 27 (59.3%) patients with septic shock. In the multivariate analysis, the identification of severe sepsis (adjusted hazard ratio [AHR] = 2.9, 95% confidence interval [CI] = 1.0-8.2, P = 0.04), septic shock (AHR = 5.7, 95% CI = 1.6-20.3, P = 0.007), and dysfunction of three or more organs (AHR = 2.9, 95% CI = 1.1-7.3, P = 0.03) increased the risk of in-hospital mortality. Adding aggregate organ dysfunction to the multivariate equation that included the sepsis category statistically significantly improved the model, but the opposite did not. Predictors of mortality were easily measurable and could be used to risk stratify critically ill patients in resource-constrained settings.