Milk samples from 340 individual goats in 34 dairy herds throughout Norway were examined for Mycobacterium avium subsp. paratuberculosis (M.a. paratuberculosis) by culture and immunomagnetic separation combined with PCR (IMS-PCR). The samples included three categories; (A) vaccinated dairy goats in herds with paratuberculosis; (B) vaccinated dairy goats in herds with no history of paratuberculosis; (C) unvaccinated goats in herds with no history of paratuberculosis. Viable M.a. paratuberculosis were not detected by culture in any sample, but 24 samples (7.1%) tested positive by IMS-PCR when the PCR products were visualised by dot blot hybridisation. PCR products from five milk samples originating from five different herds were sequenced; all showed 99% homology with the IS900 sequence from M.a. paratuberculosis.M.a. paratuberculosis were detected in all sampled categories. The percentage of IMS-PCR positive samples from herds where paratuberculosis had previously been reported was significantly lower than from herds where the infection had never been diagnosed (3.3 and 9.1%, respectively, P=0.048). Similar proportions of milk samples from vaccinated and non-vaccinated goats tested positive for the presence of M.a. paratuberculosis. Vaccinated goats older than 4 years tested positive more often than vaccinated animals less than 2 years old. Samples collected in May tested significantly more often positive than milk sampled during February-March (13.8 and 2.9%, respectively, P=0.001). This study showed that raw goats' milk in Norway might be contaminated with M.a. paratuberculosis.