Over a 3-month period, seven patients in a paediatric oncology unit developed Pseudomonas pickettii septicaemias. The outbreak was difficult to recognize since the cases occurred at widely spaced intervals and problems were experienced with the identification of the isolates. Many of the isolates were initially misidentified on the basis of a short sugar set used in the laboratory for identification of the non-fermenting Gram-negative bacilli. Moreover, the organisms had varying sensitivity patterns. The source of the organisms proved to be vials of 'sterile' distilled water which had been used for flushing the patients' indwelling Hickman lines. No further cases occurred once the use of this water was discontinued.