OBJECTIVE To identify the risk factors for nosocomial imipenem-resistant Acinetobacter baumannii (IRAB) infections. METHODS A prospective case-control study, set in an 1100-bed referral and tertiary-care hospital, of all patients who had nosocomial A. baumannii infections between January 1 and December 31, 2004. Only the first isolation of A. baumannii was considered. RESULTS IRAB was isolated from 66 (53.7%) patients and imipenem-sensitive Acinetobacter baumannii (ISAB) was isolated from 57 (46.3%) patients during the study period. The mean duration of hospital stay until A. baumannii isolation was 20.8+/-13.6 days in IRAB infections, whereas it was 15.4+/-9.4 days in ISAB infections. Of the patients, 65.2% with IRAB infections and 40.4% with ISAB infections were followed at the intensive care unit (ICU). Previous carbapenem use was present in 43.9% of the patients with IRAB and 12.3% of the patients with ISAB infection. In univariate analysis female sex, longer duration of hospital stay until infection, ICU stay, emergent surgical operation, total parenteral nutrition, having a central venous catheter, endotracheal tube, urinary catheter or nasogastric tube, previous antibiotic use, and previous administration of carbapenems were significant risk factors for IRAB infections (p<0.05). In multivariate analysis, longer duration of hospital stay until A. baumannii isolation (odds ratio (OR) 1.043; 95% confidence interval (CI) 1.003-1.084; p=0.032), previous antibiotic use (OR 5.051; 95% CI 1.004-25.396; p=0.049), and ICU stay (OR 3.100; 95% CI 1.398-6.873; p=0.005) were independently associated with imipenem resistance. CONCLUSIONS Our results suggest that the nosocomial occurrence of IRAB is strongly related to an ICU stay and duration of hospital stay, and that IRAB occurrence may be favored by the selection pressure of previously used antibiotics.