The production of extracellular galactose oxidase is limited to a few fungal species, including the important plant pathogens Fusarium graminearum and F. moniliforme. The best-studied enzyme is the one produced by the mycoparasitic fungus Cladobotryum (Dactylium) dendroides NRRL 2903. The NRRL 2903 strain was first mis-identified as Polyporus circinatus and later re-determined as Dactylium dendroides, although sporulation was never observed and the fungus was regarded as sterile. Upon growth at 25 °C, 50 rpm, in liquid medium containing 2% cellulose as the sole carbon source, and in the presence of 0·5–0·75% yeast extract, conidial production was induced in NRRL 2903, which was re-identified as Fusarium sp. The only other known commercial strain of Cladobotryum (Dactylium) dendroides able to produce galactose oxidase, ATCC 46032, also produced fusiform conidia upon growth in cellulose-containing medium, and was shown to be genetically identical to the NRRL 2903 strain. Genetic comparison with six different representative strains of Cladobotryum dendroides (teleomorph: Hypomyces rosellus), and four strains of the closely related Hypomyces aurantius, based on the analysis of the presence or absence of a homologous galactose oxidase gene ( gaoA), RAPD-PCR and RFLP analysis, confirm the distinct nature of the NRRL 2903 strain and Cladobotryum dendroides. Despite the resemblance of NRRL 2903 conidia and conidiophores to those of Fusarium chlamydosporum genetic comparison, with three different strains, suggests NRRL 2903 cannot be re-identified as F. chlamydosporum. Two of the strains, however, contain a region in their genome that is highly homologous to the galactose oxidase gene ( gaoA), and one strain exhibits extracellular galactose oxidase activity but only partial homology to the gaoA gene of NRRL 2903.