A nosocomial outbreak of Pseudomonas aeruginosa infections which occurred in the Urology Service of a large city hospital was studied. A case-control methodology was used to analyse patients' characteristics and the main risk factors of all cases with a positive culture during the period between March 1987 and March 1988. The usefulness of factor analysis in the definition of a case was examined. There were 74 infections of which 35 (47.3%), had a nosocomial origin. The outbreak took place in December 1987, with a peak incidence of infections of 10.5%, compared with a 2.2% frequency during the preceding months (P less than 0.005). Six of the nine infections occurring in that month, were caused by strains resistant to ticarcillin and gentamicin. The epidemic cases had longer hospital stays than the non-epidemic cases (P less than 0.038) and occurred more frequently in a specific area of the hospital (P less than 0.001). The odds ratio for resistance to gentamicin was 15 (P less than 0.018) and that of resistance to ticarcillin, 127 (P less than 0.0001). Our results suggest that inaccurate case definitions may produce misleading conclusions. Factor analysis appears to be a useful analytical tool when defining a case.