Although epidemiological evidence supports the fact that sheep and cattle are infected with different strains of M. paratuberculosis in Australia, this had not been confirmed. This work was designed to examine whether sheep strains in Australia are the same as are found in New Zealand, and whether the polymorphism's detected in Australian isolates of M. paratuberculosis would be useful in epidemiological investigations of Johne's disease in Australia. Seventy three isolates of M. paratuberculosis from 66 animals were examined for genetic variation using the insertion sequence IS900. The samples originated from 5 animal species in 5 Australian States and the Northern Territory. Four reference strains and an isolate from a Crohn's disease patient were also examined. The 3 restriction enzymes most commonly used from IS900 RFLP were used, and 4 additional restriction enzymes were assessed. Bst EII and Bam HI proved to be the most useful for detecting polymorphisms in Australian strains of M. paratuberculosis. In one case, isolates with a single band difference were found from different sites in the same animal. Most cattle and some other species are infected with cattle (c) strains and sheep were predominantly infected with sheep (S) strains. Thus, while the C strain infected other animal species such as alpaca, goats, a rhinoceros and sheep, there was no evidence in this study of the S strain infecting cattle. Because of the small number of C strains that had unique IS900 RFLP patterns, it is unlikely that IS900-RFLP would be useful in epidemiological investigations of Johne's disease transmission amongst cattle. Since no distinctions could be made among ovine isolates, this fingerprinting technique is unlikely to be of use in the investigation of ovine Johne's disease epidemiology other than to clarify whether infection on newly diagnosed properties was due to members of the C or S groupings.