The authors studied the impact of intensive care unit (ICU) acquired nosocomial infections on surgical patients stratified by severity of illness before acquisition of the infection. Data were analyzed from 2,122 consecutive patients admitted to a 20 bed surgical intensive care unit (SICU) from January 1, 1988 to December 31, 1988. The simplified acute physiology score (SAPS), a measure of illness severity that correlates with mortality, was calculated for all patients on their first SICU day. Ninety-seven nosocomial infections from various sites were documented in 54 patients. Patients who acquired a nosocomial infection were significantly more ill upon admission to the SICU than patients who did not acquire such an infection (control patients). Stratified by admission severity of illness, patients acquiring one or more nosocomial infections had a significantly longer SICU stay, averaging 25.3 days compared to 2.3 days in control patients (P less than 0.001). Hospital stay was also significantly increased at 59.9 days, compared to 15.0 days in control patients (P less than 0.001). However, the overall mortality rate for patients developing nosocomial infections was significantly higher than control patients only in the middle range of admission SAPS measurements. The authors found that the monthly incidence of isolates of Xanthomatous maltophilia, a multiply-resistant nosocomial organism, reflected the overall incidence of nosocomial infections in the SICU. They observed a decline in the number of new X. maltophilia isolates and nosocomial infections concomitant with the introduction of gown and glove contact isolation procedures. The authors conclude that nosocomial infections in the SICU setting are directly related to increased patient morbidity and mortality depending, in part, on severity of illness upon admission.(ABSTRACT TRUNCATED AT 250 WORDS)