Background Clusters of Candida albicans and Candida parapsilosis infections were noted intermittently in our neonatal intensive care unit (NICU), We attempted to determine whether these clusters represented single strain out-breaks or coincidental emergence of unrelated strains, Methods. A retrospective examination of the frequency of candidemia during a 9-year period, two point prevalence studies of colonization and assessment of strain relatedness of individual infant isolate during and in between clusters during a 2-year period with karyrotyping and restriction endonuclease analysis of genomic DNA (REAG). Results. C. albicans and C. parapsilosis infections emerged in a scattered pattern (1 to 2 cases every few months) with intermittent clustering of 3 cases/month, The colonization rate was 50% 5 weeks after an apparent cluster, equally distributed between C. albicans and C. parapsilosis, and 17.6% (exclusively with C. parapsilosis) 4 months after absence of invasive disease, Utilizing REAG or karyotyping singly we noted 12 and 16 DNA banding patterns, respectively, among 23 infant isolates, Few patterns were observed repeatedly over 2- to 20-month periods, implying recurrent emergence of the same strains, Combining karyotyping with REAG revealed a different epidemiologic pattern, It identified 20 distinct composites with identical composites in 3 infant pairs, All infants with identical composites were in the NICU concurrently, The frequency of strain relatedness was comparable among clustered Gases (16.7%), scattered cases (7.7%) and simultaneously colonized infants (16.7%). Conclusions. These finding;fs illustrate that Candida infections clustered periodically in our NICU but that these clusters were often caused by unrelated strains with infrequent cross-infection during and between clusters, With sub-optimal typing this pattern of emergence can be mistaken for same strain outbreaks.