The role of respiratory assistance devices and techniques in the acquisition of endemic hospital-associated pneumonia was prospectively studied in 13,086 patients over 11 months. Of these, 914 (7 percent) had a respiratory assistance device for at least 24 hours. Cultures of respirator effluent air and nebulizer fluid (taken after 24 hours), tracheostomy sites and irrigating solutions and respirometers were obtained in the 144 of 914 patients who had a respiratory assistance device for at least 72 hours. There were 108 episodes of hospital-associated pneumonia in 107 patients (0.82 percent incidence). Gram-negative organisms were associated with 70 percent of these episodes and Strep. pneumoniae with 5 percent. The risk of hospital-associated pneumonia was 0.3 percent in patients without a respiratory assistance device (35 percent of total hospital-associated pneumonia) versus (1.3 percent with endotracheal tubes and respirators (11 percent of hospital-associated pneumonia), 25 percent with tracheostomy (12 percent of hospital-associated pneumonia) and 66 percent in patients with tracheostomy and a respirator (9 percent of hospital-associated pneumonia). No case of hospital-associated pneumonia occurred in patients on respirators less than 24 hours, but the risk of hospital-associated pneumonia increased significantly after the fifth day of therapy. None of the 63 cultures of nebulizer fluid was positive. Although positive cultures of respiratory effluent, tracheal suction fluid or respirometer were not predictive of the acquisition of hospital-associated pneumonia, nine of 107 patients acquired this infection after a previously positive culture of a respiratory assistance device, and in five instances with the same organism. Since contaminated respiratory assistance devices are rarely a direct cause of hospital-associated pneumonia, routine in-use monitoring of respiratory assistance devices does not appear warranted.