To compare the utility and diagnostic accuracy of BACTEC and MGIT culture systems, a total of 41 pooled faecal samples, each containing faeces from one sheep infected with the S strain of Mycobacterium paratuberculosis and four uninfected sheep was cultured. The MGIT culture system did not support the growth of the S strain of M. paratuberculosis from faeces within the time frame of the experiments, although a laboratory adapted S strain grew slowly in MGIT provided that sufficient bacteria were inoculated. In contrast, C strain grew rapidly in MGIT. The sensitivity of culture was calculated relative to the infection status of the animals, none of which had clinical signs of ovine Johne's disease. The overall sensitivity of pooled faecal culture in the BACTEC culture system was 21.9% (95% confidence limits, 10.5-37.6), a figure dependant on the proportion of multibacillary cases. The sensitivities of the BACTEC culture system for pools containing animals with multibacillary and paucibacillary lesions were 100.0% (95% confidence limits, 47.2-100.0) and 17.8% (95% confidence limits 6.06-36.8), respectively. The contamination rate of BACTEC cultures was 9.7% compared to 14.3% for MGIT. The effect of 100 microg/ml ampicillin on the S strain of the M. paratuberculosis was examined and in both BACTEC and MGIT media it delayed growth by about 1 week. The composition of MGIT medium, particularly presence of vancomycin hydrochloride, slowed the growth of the S strain. The low content of egg yolk was considered to be another possible factor. The radiometric BACTEC culture system remains the best alternative for the culture of S strain and is recommended in circumstances where the genotype (s) of the strains present in a region/farm is either unknown or S strain.