The isolation of M. avium subsp. paratuberculosis (Map) from tissues of infected animals is both difficult and takes a long time. Moreover, the role of M. avium subsp. silvaticum (Mas) must be considered. On the other hand PCR has been proposed as an alternative to culture. This technique permits distinguishing between Mas and Map and the results can be obtained faster than in the culture. In this study we used tissue samples (ileum and lymph nodes) from experimentally infected lambs with Map and Mas, and from ewes with natural paratuberculosis infection. Both lambs and ewes had severe or minor lesions. Mycobacteria were extracted using a phase partitioning technique and the mycobacteria lysed by bead beating with zirconium beads. The DNA was extracted using standard techniques and amplified by PCR using primers specific for IS900 and IS901. Culture was made using Lowenstein-Jensen medium with and without mycobactin J and sodium pyruvate. In the experimental animals, those with severe lesions, either infected with Map or Mas were positive in culture and PCR. Among the animals with minor lesions, the PCR technique was more sensitive than culture in lambs experimentally infected with Map. In the group infected with Mas, the same animals were detected with both techniques, and in this group, some animals with clear lesions were not detected either with PCR or culture. In the natural infection, all sheep with severe lesion were both culture and PCR positive and among the sheep with minor lesions, the culture detected more animals as positive (60%) than the PCR (30%). All the animals were positive to PCR using primers to detect IS900. Moreover, the primers and the PCR technique used distinguished clearly between the animals experimentally infected with Mas or Map.