The potential use of a novel immunomagnetic PCR (IMS-PCR) technique as a rapid method to screen milk samples for the presence of Mycobacterium avium subsp. paratuberculosis (M. ptb) was assessed. Immunomagnetic separation (IMS) for M. ptb, developed at Queen's University, Belfast, was applied to milk samples prior to IS900 PCR in order to selectively concentrate any M. ptb cells present and, at the same time, separate the cells from constituents of milk likely to inhibit subsequent PCR. This increased the sensitivity of IS900 PCR. IMS-PCR sensitivity could be further increased by initial centrifugation (2500 g for 20 min) of larger volumes of milk (10 and 50 ml), and resuspension of the sediment into a 1 ml volume appropriate for IMS treatment. Following IMS, template DNA for IS900 PCR was obtained by heating the bead-cell suspension in a thermal cycler at 100 degrees C for 15 min. It was estimated that the IMS-PCR assay could detect approximately 10(3)CFU of M. ptb per 50 ml of milk (equivalent to 20 CFU/ml), whereas the minimum detection limit of direct IS900 PCR was estimated at 10(5)CFU of M. ptb per 50 ml (equivalent to 2000 CFU/ml). A blind trial was carried out in which a total of 40 spiked (10(6)CFU M. ptb) and unspiked, raw and laboratory-pasteurised milk samples were independently tested by IMS-PCR and conventional IS900 PCR. IMS-PCR correctly identified 97. 5% of milk samples (sensitivity 100%, specificity 95%), including spiked milk samples before and after laboratory-pasteurisation. One false positive result was obtained which may have resulted from carryover between samples during the IMS procedure. Conventional IS900 PCR correctly identified only 72.5% of the same 40 milk samples (sensitivity 23%, specificity 100%). IMS-PCR was also shown to be capable of detecting natural M. ptb infection in raw sheep's milk, and raw and commercially pasteurised cows' milk.