Risk factors for the selection of Stenotrophomonas maltophilia were analysed by correlating antimicrobial use and structure parameters (e.g. hospital type) with the incidence density of S. maltophilia and the percentage of S. maltophilia isolated from 39 intensive care units (ICUs). SARI (Surveillance of Antimicrobial Use and Antimicrobial Resistance in German Intensive Care Units) is a prospective unit- and laboratory-based surveillance system that collects data on the 13 most important organisms responsible for nosocomial infections. The percentage of S. maltophilia among these organisms and the number of S. maltophilia per 1000 patient-days were calculated. The data were subsequently correlated with antibiotic use density calculated in defined daily doses (DDDs) per 1000 patient-days and structure parameters. The data covered a total of 28 266 isolates and 431 351 DDDs. The antibiotic use density ranged from 427 to 2218, with the median being 1346. Over the two-year period, the median of S. maltophilia per 1000 patient-days was 1.4 (range 0-7.6). Calculation of antibiotic use and S. maltophilia per 1000 patient-days showed a significant positive correlation with the use of carbapenems, ceftazidime, glycopeptides and fluoroquinolones, as well as with total antibiotic use. In the multiple logistic regression analysis, carbapenem use and >12 ICU beds were independently and positively associated with a high number of S. maltophilia per 1000 patient-days. Benchmarking data provided for incidence densities of S. maltophilia in ICUs revealed the heterogeneous situation of the burden of S. maltophilia in individual ICUs. The multi-centre data showed that carbapenem use and >12 ICU beds were independent risk factors for the isolation of S. maltophilia.