The risk of ventilator-associated infection was studied by bacteriological monitoring of different parts of 25 ventilators functioning in a surgical ICU. Although patient tubings were not changed throughout the study, colonization of tracheal tube was found invariably to be the first phenomenon. All tracheal tubes were infected by day 9. Contamination of the expiratory limb paralleled contamination of the tracheal tube with a lag period of 2-4 days. Colonization of the inspiratory limb was a delayed and slow phenomenon which never occurred before the patient was infected. Humidifiers were found clean after the patient was disconnected. It is concluded that an adequately sterilized ventilator is not a source for hospital-acquired pulmonary infection, and that special attention to the care of tracheal tube and clinical cleanliness are of major importance to prevent infection.