Vaccination has for many years been identified as a potentially cost-effective method for the control of paratuberculosis. However, current vaccines based on whole Mycobacterium paratuberculosis bacilli mixed with an oil adjuvant have major side effects, including large lesions at the injection sites and interference with diagnostic tests for bovine tuberculosis. In this presentation we outline the steps being undertaken using molecular biological methods, to produce a new, live vaccine based on the deletion of specific genes of the paratuberculosis organism. This strategy has been successfully followed for producing new tuberculosis vaccines. Two different approaches have been used to produce avirulent mutants. The first, was to make a library of mutants in virulent M. paratuberculosis through the use of the conditionally replicating shuttle phasmid phAE94 which contains the Tn5367 transposon. The mutant library was screened using in vitro culture systems including inability to multiply in minimal media, increased temperature sensitivity, carbon source preference and altered colonial morphology, all phenotypes that have been associated with loss of virulence in other pathogenic species. The library was also screened by looking for mutants with reduced ability to survive in cultures of bovine peripheral-blood macrophages. The second approach was to inactivate specific genes through homologous recombination. The genes selected for inactivation were those which have homologues in Mycobacterium bovis and whose inactivation in M. bovis produced mutants with good vaccine efficacy against bovine tuberculosis. A crucial step in developing live vaccines is to determine the virulence of potential vaccine strains in an animal model. The loss of virulence of selected mutants has been determined by the intravenous inoculation of 10⁸ bacilli into recently weaned goats. Initial studies showed that a range of different gut tissues were colonized with moderate numbers of M. paratuberculosis one year after being infected with virulent strains. In contrast, no or very low numbers of bacilli were isolated with mutant strains. Subsequent trials have shown that virulence can be determined by examining goats after a six month infection. Immune responses were monitored by measuring gamma interferon release to Johnin PPD and will be used for selecting strains for determining their vaccine efficacy. Vaccine efficacy will be determined in a goat vaccination / challenge model. Recently weaned BALB/c mice are being investigated as a possible alternative animal for the initial screening of mutants for loss of virulence. The reduction in the number of M. paratuberculosis bacilli in spleen and liver is associated with reduced virulence.