In order to introduce specificity for Mycobacterium avium subsp. paratuberculosis prior to a phage amplification assay, various magnetic-separation approaches, involving either antibodies or peptides, were evaluated in terms of the efficiency of capture (expressed as a percentage) of M. avium subsp. paratuberculosis cells and the percentage of nonspecific binding by other Mycobacterium spp. A 50:50 mixture of MyOne Tosylactivated Dynabeads coated with the chemically synthesized M. avium subsp. paratuberculosis-specific peptides biotinylated aMp3 and biotinylated aMptD (i.e., peptide-mediated magnetic separation [PMS]) proved to be the best magnetic-separation approach for achieving 85 to 100% capture of M. avium subsp. paratuberculosis and minimal (<1%) nonspecific recovery of other Mycobacterium spp. (particularly if beads were blocked with 1% skim milk before use) from broth samples containing 10(3) to 10(4) CFU/ml. When PMS was coupled with a recently optimized phage amplification assay and used to detect M. avium subsp. paratuberculosis in 50-ml volumes of spiked milk, the mean 50% limit of detection (LOD(50)) was 14.4 PFU/50 ml of milk (equivalent to 0.3 PFU/ml). This PMS-phage assay represents a novel, rapid method for the detection and enumeration of viable M. avium subsp. paratuberculosis organisms in milk, and potentially other sample matrices, with results available within 48 h.