The contribution of ventilator circuit bacterial contamination to the occurrence of ventilator-associated pneumonia remains controversial. In a previous study, we found that the incidence of pneumonia was identical with ventilator circuit changes every 48 h and with no ventilator circuit changes. The present study prospectively assessed whether keeping ventilator circuits clean with a heat and moisture exchanger exhibiting antimicrobial barrier properties affects patient colonization and the incidence of nosocomial pneumonia in patients receiving mechanical ventilation for more than 48 h. Consecutive patients were randomly allocated to humidification with either a heat and moisture exchanger (Group 1, n = 61) or a heated humidifier (Group 2, n = 70). In both groups, no circuit changes were performed throughout ventilatory support. Duration of mechanical ventilation was identical in both groups (10 +/- 8.6 d (range: 2 to 47) in Group 1 and 12.5 +/- 14.2 d [range: 2 to 85] in Group 2). The incidence of pneumonia (positive quantitative culture of protected brush specimen) was similar in both groups (6/61 and 8/70 in Groups 1 and 2, respectively; p = 0.8), as was duration of ventilation prior to pneumonia (9 +/- 5.9 versus 8.2 +/- 5.7 d; p = 0.8). Ventilator tubing contamination was considerably reduced with the use of a heat and moisture exchanger. In contrast, bacterial colonization of the pharynx and trachea was identical in both groups. These results suggest that circuit colonization plays little or no role in the occurrence of ventilator-associated pneumonia, provided usual maintenance precautions are applied.(ABSTRACT TRUNCATED AT 250 WORDS)