Within six months of opening of the new Dublin Dental Hospital in September 1998, areas of corrosion were observed on many of the baseplates of the hospital's 103 dental chair units (DCUs) at the site of attachment of the suction hoses. The corroded areas were heavily contaminated with Pseudomonas spp. and related genera posing a risk of cross-infection, particularly for immunocompromised patients. These species were used as marker organisms to investigate the source of the contamination. P. aeruginosa was the predominant species recovered from 41 selected DCU baseplates (61% prevalence), whereas P. putida (46% prevalence) and P. aeruginosa (43% prevalence) were predominant at the attachment ends of 37 selected high-volume suction hoses. Forty-one selected isolates of P. aeruginosa from 13 DCU baseplates, 16 high-volume suction hoses and 12 coarse filter housings (another suction system site) from 19 separate DCUs were serotyped to determine the similarity of isolates at each site. The majority of isolates (68.3%) belonged to serotype O:10, while the remainder belonged to serotypes O:6 (7.3%), O:11 (7.3%), O:14 (9.8%) and O:5/O:16 (7.3%). Of the isolates from DCU baseplates, additional isolates with the same serotype were recovered from other suction system sites in 10/13 (77%) cases. Isolates of only one serotype were recovered from each of the 19 DCUs investigated. Forty-one serotyped isolates were also subject to computer-assisted analysis of SpeI-generated DNA fingerprint profiles, and similarity coefficient (S(AB)s) values were calculated for each pairwise combination of isolate profiles. The data obtained showed that the isolates consisted of two distinct main populations, each containing separate clades corresponding to specific serotypes. Serotype O:6 (three isolates), O:11 (three isolates) and O:5/O:16 (three isolates) belonged to a single strain in each case. Serotypes O:14 (four isolates) and O:10 (28 isolates) belonged to two strains in each case. The two serotype O:10 strains, termed fingerprint groups I (four isolates from three DCUs) and II (24 isolates from 10 DCUs), were the most distantly related of all the strains identified. These findings demonstrated that the hospital DCUs had become colonized with a small number of P. aeruginosa strains, one of which (serotype O:10, fingerprint group II) predominated. These results also confirmed that DCU baseplate contamination was most likely to be due to leakage from suction system hoses at the baseplate attachment sites, probably due to loosening during use. Replacement hose connectors that firmly retained the suction hoses in the attachment sites so that they could not be loosened by movement of the suction hoses solved this problem, and eliminated further contamination of the DCU baseplates.