Prospective studies were conducted for nosocomial Pseudomonas aeruginosa infections from February 1, 1994, to October 30,1995. Of 97 P aeruginosa isolates from 97 patients, 35 were resistant to ceftazidime. Logistic regression revealed previous cephalosporin or piperacillin use as independent risk factors for nosocomial ceftazidime-resistant P aeruginosa infection. Pulsed-field gel electrophoresis revealed that four nosocomial ceftazidime-resistant P aeruginosa infections were caused by cross-infection, probably through medical personnel.