Very little was known on the epidemiology of ovine paratuberculosis in the North-East of Portugal until a recent serological study in the Trás-os-Montes e Alto Douro region showed a high flock and moderate sheep prevalence. In order to improve and confirm the seroprevalence data and to explore other diagnostic alternatives, a study based in detection in blood of MAP DNA was carried out. A total of 150 flocks were included in this survey. In order to limit the costs for each flock a set of five clinically healthy sheep and another one of five clinically suspect paratuberculosis sheep were sampled. In healthy sheep only blood were taken to pooled PCR and individual ELISA. From clinically suspected sheep both faeces and blood were taken. Faeces were submitted to pooled PCR and isolation on LJ and Middlebrook 7H11 and blood to absorbed individual ELISA and pooled PCR. One thousand and five hundred sheep were examined in total. Results with each method were compared to each other in order to estimate complementary sensitivity, agreement and cost efficiency. PCR detected MAP DNA in 56 (18.7%) pool samples. The overall individual prevalence of MAP was calculated to range from 6.4 to 15.4%. MAP was found in 20.7% pooled samples from apparently healthy animals and in 16.7% pooled samples from suspect animals. Fecal isolation and PCR showed an overall prevalence of 0.9% and 1.9% respectively. Agreement between pooled methods was generally low, but for healthy versus suspect blood PCR which was moderate. Healthy and suspect pooled ELISA showed a similar average estimate for flock and individual prevalence but costs were much lower. The results of this study also suggest that the health status is not closely related to the true level of infection with MAP.