In a retrospective study, children with cystic fibrosis who were colonised with Pseudomonas cepacia were compared with a control group who were colonised with Pseudomonas maltophilia. Out of 216 children with cystic fibrosis seen between 1983 and 1990, P cepacia was recovered from 13 (median age at colonisation 12.2 years) and P maltophilia from 23 (median age at first colonisation 6.1 years), and both organisms were recovered in five cases. With the exception of two patients with P cepacia in whom no other pathogens were found, all the patients with P cepacia or P maltophilia had co-colonisation with Pseudomonas aeruginosa. The lack of spread of P cepacia to siblings with cystic fibrosis, and the relative lack of inpatient contact between colonised and uncolonised patients suggest that cross infection is not the sole route whereby patients with cystic fibrosis become infected, but the possibility of cross infection cannot be excluded from our data. Three patients with P cepacia died, but two of these had shown appreciable respiratory deterioration before colonisation with P cepacia; there was no evidence of unexpected deterioration in the remainder or in the controls with P maltophilia. By 1990, the prevalence of P cepacia was 9/133 (7%) and that of P maltophilia was 13/133 (10%), but it was impossible to determine to what extent this increase was due to the introduction of the routine use of selective media. Further studies are required to establish whether patients with and without P cepacia should be segregated.