In a randomized clinical trial the prophylactic effects of locally administered antimicrobials on quantitative colonization and respiratory infections were studied in intubated patients with an expected period of mechanical ventilation of greater than 6 days. Nineteen patients received 50 mg of polymyxin B and 80 mg of gentamicin distributed among nose, oropharynx and stomach at 6-h intervals, as well as 300 mg of amphotericin B in the oropharynx. Twenty untreated patients served as controls. In the control group colonization by respiratory pathogens was more common (oropharynx 19 vs 6 patients (p less than 0.001); trachea 19 vs 11 (p less than 0.01)), and the number as well as the count of the colonizing species was usually higher. Fourteen patients of the control group developed respiratory infections, including nine cases of pneumonia, as compared to four patients with prophylaxis, including one case of pneumonia (p less than 0.01). Pneumonia-associated deaths were prevented with prophylaxis; however, the overall mortality remained unchanged. Respiratory infections in the prophylaxis group were associated with organisms resistant to the agents used, but the overall occurrence of resistance was not increased, as compared to the control group. We conclude that unrestrained upper airway colonization by respiratory pathogens and respiratory tract infection were causally related. Local antimicrobial prophylaxis proved to be a highly effective strategy for the prevention of potentially life-threatening pneumonias in critically ill patients, but in the present study the host setting appeared to be the major determinant of outcome.