In an earlier study, nosocomial infections were reported to occur in one third of patients who died in the hospital. Lower respiratory tract infections appeared to be causally related to death, whereas urinary tract infections were not. A case-control study was therefore conducted to begin to explore epidemiologically these causal suppositions. By definition, all 100 patients in the case group died in the hospital and all 100 patients in the control group survived hospitalization. A patient in the control group was matched to a patient in the case group for age, sex, service, primary discharge diagnosis, and severity of primary diagnosis. Secondary discharge diagnoses were comparable in both groups. Nosocomial infections were present in 33 percent of patients in the case group and 13 percent of patients in the control group (p < 0.001). Lower respiratory tract nosocomial infections occurred in 18 percent of patients in the case group and 4 percent of patients in the control group (p < 0.005). Urinary tract nosocomial infections occurred in 6 percent of patients in the case group and 8 percent of patients in the control group (p = NS). Presence of a terminal prognosis on admission was found more often in case group patients than in control group patients (76 percent versus 11 percent, p < 0.001). Nosocomial infections appeared to favor a fatal outcome only in those whose condition was not terminal on admission (p < 0.05). In those whose condition was terminal on admission, the incidence of infection was equal in those who survived and in those who died. In conclusion, in these two groups well matched by many criteria, differences in prognosis on admission probably accounted for the major differences in survival. Nosocomial infections may affect outcome in those whose condition is not terminal on admission.