To determine the epidemiology and risk factors for carbapenem-resistant Pseudomonas aeruginosa (CR-PA) digestive tract colonization, weekly rectal and pharyngeal swabs were obtained in two serial incidence surveys (266 patients). Forty-two (16%) patients were CR-PA colonized (12 [29%] on admission and 30 [71%] in intensive care units). Pulsed-field gel electrophoresis showed extensive clonal diversity, although one specific clone (type B) was isolated from 11 patients. The presence of similar genotypes of CR-PA colonizing 30% of the CR-PA-colonized patients suggests the occurrence of cross-colonization; in addition, 10 pairs of carbapenem-susceptible P. aeruginosa (CS-PA) and subsequent CR-PA strains isolated from the same patients were found to be clonally identical and were considered to have been endogenously acquired (33%). All endogenously acquired CR-PA strains were isolated after exposure to a carbapenem, and 80% showed a phenotype of imipenem resistance (IR pattern) alone, while 67% of the CR-PA strains acquired by cross-transmission exhibited a multiresistant (MR) phenotype, with previous carbapenem exposure in 44%. Logistic regression analysis identified severity of acute illness (odds ratio [OR], 1.0; 95% confidence interval [CI], 1.0 to 1.1), prior carbapenem use (OR, 7.8; 95% CI, 1.7 to 35.3), and prior use of fluoroquinolones (OR, 11.0; 95% CI, 1.7 to 67.9) as independent risk factors for CR-PA digestive tract colonization. Overall, the local epidemiology of CR-PA digestive tract colonization was characterized by polyclonal endemicity with phenotype patterns of IR and MR divided evenly between patients. Restricting the use of particular agents, such as carbapenems and fluoroquinolones, should be considered advisable to minimize the problem of this antibiotic resistance. However, the possible risk for development of collateral unexpected bacterial resistance patterns should be accurately monitored.