the purpose of the data was to explore the performance of the bacterial filters (BF) on the incidence of ventilator-associated pneumonia.totally, 206 patients considered to require MV for more than 48h on admission were enrolled. 86 were assigned to BF group, 120 were assigned to NBF group; 24 patients (11 in BF group, 13 in NBF group) were secondarily excluded for the duration of intubation was less than 48h. one patient was excluded for the lack of chest radiograph before intubation. The characteristics of the remaining patients (75 in BF group and 106 in NBF group) were analyzed at last.The characteristics of the two group were shoqed in table1.Univariable logistic analysis was applied to demonstrate the factors contributing to the development of VAP. The results were showed in the Table 2.The retrospective subgroup analysis results were showed in Table 3. cox proportional hazards regression model was showed in table4. in clusion, we found that bacterial filters placed on the inspiratory limbs of breathing circuit did not reduce VAP in ICU patients ventilated for more than 48h when the gas was humidified with heated water-bath humidifiers with double heater-wire circuit. Interestingly, the filter might protect the patients from occurrence of VAP when ventilated for not long enough.