Prospective surveillance for nosocomial infections was performed for a five-year admission cohort (1980-1984) at North Carolina Memorial Hospital. One or more nosocomial infections developed in 2,662 patients (2.6%) from 102,206 patients at risk; greater than or equal to 2 nosocomial infections developed in 775 of these 2,662 patients (29.1%), and greater than or equal to 3 nosocomial infections in 304 of 775 patients with greater than or equal to 2 infections (39.2%). Hospital stay was significantly prolonged for infected compared with never-infected patients (38.1 vs. 7.9 days, p less than 0.0001) and for multiply-infected versus once-infected patients (57.9 vs. 30.0 days, p less than 0.0001). Total nosocomial infections numbered 4,031 with 2,144 multiple infections (53%); the average number of nosocomial infections per infected patient was 1.5 (4,031 infections in 2,662 patients). Among all nosocomial infections, 64% of bacteremias, 55% of respiratory infections, 55% of surgical wound infections, and 40% of urinary tract infections occurred in patients with multiple nosocomial infections. Surgical patients had 56% of multiple infections. Intensive care unit patients had significantly more multiple infections than non-intensive care unit patients. Nosocomial infections in intensive care unit patients were 71% multiple nosocomial infections. The probability of developing multiple infections was 11 times greater after the first infection occurred. This emphasizes the need to prevent initial nosocomial infections and to identify risk factors for multiple nosocomial infections. Determining risk factors for multiple nosocomial infections could focus infection control efforts on a subpopulation of patients who acquire over 50% of all nosocomial infections and who have significantly prolonged and costly hospital stays.