BACKGROUND: We evaluated the usefulness of a newly developed molecular typing method of infrequent restriction site polymerase chain reaction (IRS-PCR) as an epidemiological DNA fingerprinting tool for Candida tropicalis. METHODS: Thirty-two strains of C. tropicalis comprising eight sporadic strains and 24 clonal strains belonging to six clones, of which clonal type were previously confirmed by pulsed-field gel electrophoresis (PFGE), were tested by IRS-PCR to evaluate the usefulness of this technique. Twenty strains of Candida species, including C. glabrata, C. krusei, C. albicans, and C. parapsilosis, were also tested to assess the ability of IRS-PCR to discriminate among species of Candida. RESULTS: Using the IRS-PCR assay, sporadic strains of C. tropicalis could not be differentiated from clonal strains. Most strains belonging to the same clones were classified as different IRS-PCR types or clusters, and some different sporadic strains were classified as the same IRS-PCR types. When pattern variation was examined for different strains of C. tropicalis using IRS-PCR, pairwise similarity measured by the Dice coefficient was 75.4~100%. In contrast, pairwise similarity among isolates of five different species of Candida was 25~69.2%. Therefore, five different species of Candida were easily differentiated. CONCLUSION: The IRS-PCR typing assay appears to be an inadequate tool for the epidemiological typing of C. tropicalis, because the typing result of IRSPCR is not comparable to that of PFGE. To our knowledge, this is the first evaluation study for IRSPCR as an epidemiological typing tool for C. tropicalis.