BACKGROUND: This analysis is part of a multicenter study conducted in Israel to evaluate survival of critically ill patients treated in and out of intensive care units (ICUs). OBJECTIVE: To assess the role of infection on 30-day survival among critically ill patients hospitalized in ICUs and regular wards. DESIGN: All adult inpatients were screened on four rounds for patients meeting ICU admission criteria. Retrospective chart review was used to detect presence and type of infection. Mortality was ascertained from day of meeting study criteria to 30 days thereafter. ANALYSIS: The effect of infection on mortality among patients, treated in and out of the ICU, was compared using Kaplan Meier survival curves. Multivariate Cox models were constructed to adjust interdepartmental comparisons for case-mix differences. RESULTS: Of 641 critically ill patients identified, 36.8% already had an infection on day 0. An additional 40.2% subsequently developed a new infection during the follow-up period, ranging from 64.6% in the ICU to 31.5% in regular wards (p < .001). Resistant infections were more prevalent in ICUs. Infection was independently associated with an increase in mortality, regardless of whether the patient was admitted to the ICU. There was no difference in the adjusted risk of mortality associated with an infection diagnosed on day 0 vs. an infection diagnosed later. Risk of dying was similar in resistant and nonresistant infections. Adjusting for infections, survival of ICU patients was better relative to patients in regular wards (adjusted hazard ratio = 0.7). Among the different types of infection, risk of mortality from pneumonia was significantly lower in ICUs relative to regular wards. There was a protective effect in ICUs among noninfected patients. CONCLUSION: The risk of acquiring a new infection is greater in the ICU. However, risk of mortality among ICU patients was lower for the most serious infections and for those without any infection.