A duplex polymerase chain reaction (PCR)-hybridization assay based on Mycobacterium avium subsp. paratuberculosis (MAP)-specific IS900 integration sites was used to evaluate two mycobacterial recovery methods from bovine feces: a direct-dilution-centrifugation method and a C(18)-carboxypropylbetaine (CB-18)-based method. All MAP PCR results were confirmed for absence of inhibitors using a novel PCR system based on the rpoB gene of plant chloroplasts as an internal control. The detection limits of both MAP recovery methods when coupled with PCR were determined to be between 100 and 1000 organisms. Using culture as a 'gold standard' PCR following the direct-dilution-centrifugation protocol was 92.6% sensitive and 83.7% specific, whereas PCR following the CB-18 method was 100% sensitive and 53.5% specific. Both methods were 100% specific when 60 'true' negatives from two uninfected herds were tested. Both the CB-18 and direct processing methods coupled with a target-specific amplification technique may provide greater sensitivity to diagnose subclinical animals as they were able to detect more positives, on samples derived from infected herds, than conventional culture methods; however, more extensive investigation and follow-up of suspect animals will be required to fully validate the MAP recovery and molecular detection protocols described.