Because of a high prevalence of Pseudomonas aeruginosa infections, we conducted an epidemiological study to assess the need for systematic surveillance, as well as the value of applying barrier precautions toP. aeruginosa carriers. From July 1997 to February 1998, we conducted a prospective cohort study in an 18-bed medical intensive care unit (ICU), which is part of the infectious diseases department in a 1200-bed tertiary-care teaching hospital. Rectal and oropharyngeal swabs were obtained on admission and twice weekly. Acquired strains were genotypically characterized by pulsed-field gel electrophoresis (PFGE). A risk factor analysis for carriage, colonization and infection was performed. Among 269 eligible patients, 116 (43%) were P. aeruginosa carriers, with 46 (17%) detected on admission and 70 (26%) who acquired carriage during their stay in ICU. Among these 70 patients, 29 became colonized (N=13) or developed infection (N=16). Conversely, in the 121 patients who remained free of carriage, no colonization or infection were detected. Genotyping analysis using PFGE was performed for 81/85 (95%) acquired strains in 67 patients. The same genotype I was observed for 58/81 (70%) of these strains issued from 47 patients, and a distinct genotype II affected two other patients (three strains). The last 20 strains were not genetically related. In a multivariate model, mechanical ventilation was associated with the acquisition of P. aeruginosa carriage. Antibiotics ineffective against P. aeruginosa significantly increased the risk of colonization or infection in ICU. Although several recent studies concluded that endogenous sources account for the majority of P. aeruginosa colonizations or infections, we conclude that epidemiology may vary according to the ICU, and that cross-colonization (i.e., exogenous source) may occur and warrant reinforced barrier precautions.