We enrolled 382 patients at two hospitals in Uganda in a prospective observational study of severe sepsis. Because artemisinins improve survival in murine sepsis models, we performed a post hoc analysis of the association between the use of artemether-lumefantrine (A-L) and mortality in patients with or without malaria. In patients with negative malaria smears (N = 328 of 379), Kaplan-Meier curves revealed decreased combined inpatient and 30-day mortality among patients receiving A-L versus those who did not (20.6%, SE = 10.6 versus 48.8%, SE = 3.2; Log rank chi(2) = 3.93, P = 0.048). The decrease in mortality associated with A-L was maintained in the most clinically ill patients determined by Karnofsky Performance Scores <or= 50 (16.7%, SE = 15.2 versus 58.3%, SE = 3.7; Log rank chi(2) 3.94, P = 0.041). Research into the properties of A-L is needed to improve treatment of sepsis without compromising malarial susceptibility.