Background: We have evaluated a sensitive screening assay for Mycobacterium tuberculosis (MTB) complex organisms and a specific assay for detecting Mycobacterium bovis DNA in lymph nodes taken from cattle with evidence of bovine tuberculosis. Underlying these series of experiments was the need for a versatile DNA extraction protocol which could handle tissue samples and with the potential for automation. The target for the screening assay was the multi-copy insertion element IS/08/, present in 6 copies in the MTB complex. For confirmation of M. bovis we used primers flanking a specific deletion in the genome of M. bovis known as region of difference 4 (RD4). The sensitivity and specificity of these PCRs has been tested on genomic DNA from MTB complex reference strains, mycobacteria other than tuberculosis (MOTT), spiked samples and on clinical material. Results: The minimum detection limits of the IS/08/method was < I genome copy and for the RD4 PCR was 5 genome copies. Both methods can be readily adapted for quantitative PCR with the use of SYBR Green intercalating dye on the RotorGene 3000 platform (Corbett Research). Initial testing of field samples of bovine lymph nodes with visible lesions (VL, n = 109) highlighted two shortfalls of the molecular approach. Firstly, comparison of IS/08/PCR with the "gold standard" of culture showed a sensitivity of approximately 70%. The sensitivity of the RD4 PCR method was 50%. Secondly, the success rate of spoligotyping applied directly to clinical material was 51% compared with cultures. A series of further experiments indicated that the discrepancy between sensitivity of detection found with purified mycobacterial DNA and direct testing of field samples was due to limited mycobacterial DNA recovery from tissue homogenates rather than PCR inhibition. The resilient mycobacterial cell wall, the presence of tissue debris and the paucibacillary nature of some cattle VL tissue may all contribute to this observation. Any of these factors may restrict application of other more discriminant typing methods. A simple means of increasing the efficiency of mycobacterial DNA recovery was assessed using a further pool of 95 cattle VL. Following modification of the extraction protocol, detection rate with the IS/08/and RD4 methods increased to 91% and 59% respectively. Conclusion: The IS/08/PCR is a realistic screening method for rapid identification of positive cases but the sensitivity of single copy methods, like RD4 and also of spoligotyping will need to be improved to make these applicable for direct testing of tissue extracts.