A retrospective study was done to determine the epidemiology of infection and/or colonization due to Pseudomonas aeruginosa in a Brazilian general hospital. In 1966, 1968, and 1969, there were only two instances where probable cross-contamination was shown; the remaining isolates were unrelated. In late 1971 the hospital experienced a marked increase in P. aeruginosa isolation. Contaminated dextrose solutions used in the infant feeding were the apparent cause of the problem which occurred in the premature and special care nurseries. A contaminated oxygen humidifying bottle was the source of a different outbreak in surgery. There was also evidence in four instances that cross-infection and/or contamination had occurred. Pyocin and serological typing revealed that many strains were involved and led to a clear understanding of the complex epidemiological relationships among all the strains.