Definitive diagnosis of Johne's disease in ruminants depends on confirming the presence of the causative bacterium, Mycobacterium avium subsp. paratuberculosis, in tissues of the host. This is readily achieved in most ruminant species by culture. However, culture of clinical specimens from sheep in many countries has been unrewarding. Such a culture from sheep was achieved recently in Australia by using a radiometric culture medium. The aims of the present study were to evaluate the culture of M. avium subsp. paratuberculosis from sheep by using modified BACTEC 12B radiometric medium, to determine the sensitivity of culture in relation to histopathology, and to evaluate a range of solid media. Culture of M. avium subsp. paratuberculosis from sheep with Johne's disease is a sensitive method of diagnosis: intestinal tissues from all 43 animals with multibacillary disease and all 22 animals with paucibacillary disease were culture positive, while 98% of feces from 53 animals with multibacillary disease and 48% of feces from 31 animals with paucibacillary disease were culture positive. Of sheep without histological evidence of Johne's disease from infected flocks, intestinal tissue from 32% of 41 were culture positive, while feces from 17% of 41 were culture positive. Consequently, culture is recommended as the "gold standard" test for detection of ovine Johne's disease. Of the wide range of solid media that were evaluated, only modified Middlebrook 7H10 and 7H11 agars, which were very similar in composition to modified BACTEC 12B medium, yielded growth of ovine strains of M. avium subsp. paratuberculosis. The sensitivity of detection of M. avium subsp. paratuberculosis on solid media was slightly lower than that in modified BACTEC 12B radiometric medium. Both egg yolk and mycobactin J were essential additives for growth of ovine strains of M. avium subsp. paratuberculosis in both liquid and solid media.