Dilution rates for pooled faecal culture (PFC) and direct IS900 polymerase chain reaction (D-PCR) tests were evaluated on faecal samples from infected cows mixed with uninfected faeces in dilutions from 1 in 5 to 1 in 50. PFC was performed by radiometric culture, with confirmation by IS900 PCR and restriction endonuclease analysis (PCR/REA) on growth, and by mycobactin dependency testing on solid medium. Using 37 culture positive faecal samples from 12 subclinical cows, 83.8% and 94.6% of samples were confirmed positive in the PFC assay at dilutions of 1 in 50 and 1 in 30, respectively. Lower dilutions (1 in 5 to 1 in 20) provided only marginally better sensitivity, and confirmation of PFC growth by PCR/REA was significantly more sensitive than mycobactin dependency. D-PCR had significantly lower sensitivity than PFC confirmed by PCR/REA, with pools of 1 in 50, 30, 10 and 5 yielding positive results in 64.9%, 70.3%, 78.4% and 83.8% of samples, respectively. Cattle considered to be shedding 1.5 x 10(6) viable M. avium subsp. paratuberculosis (Map)/g faeces (on the basis of estimated losses in processing and growth rates in radiometric broth) were positive at dilutions up to 1 in 50 in the PFC and D-PCR. Those shedding 5 x 10(5) viable Map/g were positive in the PFC at dilutions up to 1 in 40, but required a 1 in 10 dilution or less for D-PCR. The results suggest that for cattle shedding relatively high concentrations of Map in faeces (>2 x 10(5) viable Map/g), maximal dilutions of 1 in 30 for PFC and 1 in 10 for D-PCR would be applicable.