Summary Over a period of 20 mth all the patients with tracheostomy in an intensive care unit for major injuries were examined frequently, often daily, for the presence of Pseudomonas aeruginosa in the tracheal mucus. During the same period the hands of staff and the inanimate environment in the Unit were examined weekly for the presence of Ps. aeruginosa. Numerous isolates of Ps. aeruginosa from these sources were typed by phage and serological methods. Eight outbreaks occurred in which patients had infection with Ps. aeruginosa in the trachea, and there were seven periods between outbreaks when Ps. aeruginosa was not found in samples from tracheostomies. Strains of Ps. aeruginosa isolated in successive outbreaks were of different types. Strains isolated from the environment were usually of types already present in tracheostomies. More rarely strains of types already present in the environment were subsequently found in tracheostomies. There was no endemic infecting strain in patients, though a single type was isolated from sinks, as well as strains of other types that were transient, throughout almost the whole period of the study. This strain from sinks caused infection in patients during four of the eight outbreaks, but was not the first strain to cause infection except in the first outbreak. Of the numerous sources from which Ps. aeruginosa was isolated, and from which patients might become infected, the hands of nurses, physiotherapists and other members of staff seemed to present a special hazard of transferring infection. Other potentially important sources were plastic washing bowls, food and food mixers, and suction apparatus. Many sites commonly associated with Ps. aeruginosa contamination, e.g., nailbrushes, dishcloths, floors and floor mops, sinks and sink mops, were found to be contaminated. Samples of air occasionally yielded a few Ps. aeruginosa. Respiratory ventilators were used on some of the patients who became infected with Ps. aeruginosa, but the evidence suggested that these were not an important source of infection. Blower-humidifiers sometimes became contaminated, but experimental study of this equipment suggested that it was not a likely source of infection. Self-infection may have occurred in one patient. A number of recommendations were made, including the use of disposable plastic gloves for a wide range of nursing procedures, the supply of individual plastic washing bowls disinfected daily, and improvements in the arrangement for suction. Sometimes infection seemed to be due to lapses in technique caused by ignorance or by pressure of work. Prevention of infection of tracheostomies remains a challenge to vigilance and discipline.