To develop an experimental model of ventilator-acquired pneumonia (VAP), we investigated whether healthy piglets could develop endogenously acquired pulmonary infection as a result of prolonged mechanical ventilation (MV). Thirty-three piglets underwent MV with anesthesia, analgesia, and paralysis produced by continuous infusion of midazolam, fentanyl, and pancuronium bromide. Ten animals received antibioprophylaxis with ceftriaxone (ATB group) and 23 received no antibiotics (control group). Eighteen control animals and 9 ceftriaxone-treated animals completed the 4-day study protocol. The presence of pneumonia on day 4 was ascertained by multiple pulmonary biopsy specimens, processed for microscopic examination and quantitative cultures. The anesthetic regimen provided satisfactory electrolyte balance and cardiovascular stability. Under these circumstances, 17 of 18 animals and 4 of 9 animals developed VAP in the control and the ATB groups, respectively. Lesions of different grades of severity were unevenly distributed through both lungs with a predominance and a higher severity in dependent lung segments. Noninfectious lesions frequently associated with VAP in humans were not observed. Pneumonia was usually polymicrobial with a predominance of Gram-negative organisms. Most of the causative organisms originated from the oropharynx. Histologic lesions and lung bacterial concentrations were less in the ATB group than in control animals. We then investigated the effects of intrabronchial challenge with bacterial pathogens in the absence of MV. Intrabronchial bacterial inoculation resulted in the development of pneumonia that spontaneously resolved even when using very highly titrated inocula. Therefore, MV seems to be the main predisposing factor in the development of pneumonia in this model. This model that resembles human VAP in its histologic, bacteriologic, and pathogenic aspects may be useful to further study pathogenesis, diagnosis, prevention, and therapy of VAP.