To test the hypothesis that a new program of antibiotic strategy control can minimize the incidence of ventilator-associated pneumonia (VAP) caused by potentially antibiotic-resistant microorganisms, we performed a prospective before-after study in 3, 455 patients admitted to a single intensive care unit over a 4-yr period. Regarding the bacterial ecology and the increasing antimicrobial resistance in our medical intensive care unit (MICU), we decided to vary our choice of empiric and therapeutic antibiotic treatment, with a supervised rotation, and a restricted use of ceftazidime and ciprofloxacin, which were widely prescribed before this scheduled change. For all patients, VAP was diagnosed based on the results of quantitative culture of bronchoalveolar lavage specimens (>/= 10(4) cfu/ml). We studied 1,044 and 1,022 patients requiring more than 48 h of mechanical ventilation (MV), respectively, in the before-period (2 yr: 1995-1996) and the after-period (2 yr: 1997-1998). We observed a decrease from 231 consecutive episodes of VAP in the before-period to 161 episodes of VAP in the after-period (p < 0.01), particularly for VAP occurring before 7 d of MV. The total number of potentially antibiotic-resistant gram-negative bacilli responsible for VAP such as Pseudomonas aeruginosa, Burkholderia cepacia, Steno-trophomonas maltophilia, and Acinetobacter baumanii decreased from 140 to 79 isolated bacilli. The susceptibilities of these bacteria to the antibiotics regimen increased significantly, especially for P. aeruginosa and B. cepacia. The percentage of methicillin-sensitive Staphylococcus aureus increased significantly from 40% to 60% of S. aureus responsible for VAP. These results suggest that a new strategy of antibiotics use could be an efficient means to reduce the incidence of VAP caused by antibiotic-resistant bacteria. Nevertheless, further studies are needed to validate these data.