A total of 1,001 consecutive episodes of nosocomial pneumonia in 901 patients was identified by routine surveillance at the University of Virginia Medical Center between 1979 and 1983 (8.6 episodes/1,000 admissions). When only initial episodes were examined, 890 patients comprised the study sample. The overall case fatality rate was 30%. Stepwise logistic regression indicated that time from admission to pneumonia (p = 0.0006), age (p less than 0.0001), prior use of mechanical ventilation (p = 0.0032), and neoplastic disease (p = 0.0062) were associated with mortality. Multiple regression analysis indicated that the factors associated with increased length of hospitalization included posttracheostomy status (p = 0.0001), prior mechanical ventilation (p = 0.0001), immunosuppressive or leukopenic status (p = 0.0009), nasogastric intubation (p = 0.0003), and prior bacteremia (p = 0.0127). A sampled, individually matched cohort study (n = 74 pairs) was conducted to determine the proportion of mortality in cases that was attributable to infections (33%) and to determine excess hospital stay (seven days) among the patients with nosocomial pneumonia. Excess stay was statistically significant (p less than 0.0001), but proportional mortality was only marginally significant (p = 0.0892). Our findings suggest that nosocomial pneumonia accounts for approximately 33% of the crude mortality and contributes significantly to the economic burden associated with prolonged hospitalization.