Objective: To assess the impact of nosocomial infection on the mortality rate in an intensive care unit (ICU). Design: Prospective cohort study. Setting: The ICU of the University of Granada Hospital in Spain. Patients: All patients (n=279) admitted for ≥48 hrs at the ICU between December 1986 and April 1988. Measurements: Nosocomial infections were diagnosed according to Study on the Efficacy of Nosocomial Infection Control (SENIC) and Centers for Disease Control criteria. Patient severity on admission (using Acute Physiology and Chronic Health Evaluation (APACHE II) and Therapeutic Intensity Scoring Systems were also used. Results: Mortality risk was 2.48 times higher in patients with a nosocomial infection than in noninfected patients. Relative risk of mortality in nosocomially infected patients was higher in young and less severely ill patients, in those patients with respiratory diseases, and in those patients with longer ICU stays. Logistic stepwise regression analysis, adjusting for several confounding factors (affected organ system, APACHE II score, and therapeutic intensity), showed that the risk of death in nosocomially infected patients was 2.1 times greater (95% confidence interval = 1.0 to 4.41) than in patients without such infection. Conclusions: Nosocomial infection increases the risk of death. The effect is stronger in younger and less severely ill patients. (Crit Care Med 1994; 22:55‐60)