The recovery and isolation of M. paratuberculosis in faeces represent one of the major limits in the development and application of control programs in dairy herds. Up to now, different protocols have been proposed to isolate the pathogen from the faeces, but none of them showed to be fully satisfactory. The aim of our work was to compare different extraction protocols and commercial kits suitable for further application of PCR technique. Composite samples of negative faeces were contaminated with a single strain of M. paratuberculosis taken from an agar-plate and suspended in 1 ml sterile water, positive samples were obtained mixing 1g of faeces and 250 ml of bacteria suspension and stored at -20°C. A nested PCR protocol based on primer suggested by Linsby et al. (1994), was applied to the extracts from the different protocols. They were both experimental protocols and commercially available kits, among these latter ones some suggested for DNA extraction from plant and mouse tail, the other specifically developed for detecting M. paratuberculosis by PCR from bovine faeces. The results showed that only this latter method gave a positive nested-PCR amplification. The preliminary sampling of faeces taken from seropositive cows confirmed these results and therefore suggests that this protocol could be applied in field in the programs aimed to eradicate the disease from dairy herds.