Paratuberculosis is a chronic infection of economic importance to the dairy industry. The infection may be latent for years, which makes diagnostic misclassification a general challenge. The objective of this study was to identify the spatial pattern in infection prevalence, when results were adjusted for covariate information and diagnostic misclassification. Furthermore, we compared the estimated spatial pattern with the spatial pattern obtained without adjustment for misclassification. The study included 1242 herds in 2009 and 979 herds in 2013. The within-herd prevalence was modelled using a hierarchical logistic regression model and included a spatial component modelled by a continuous Gaussian field. The Stochastic Partial Differential Equation (SPDE) approach and Integrated Nested Laplace Approximation (INLA) were used for Bayesian inference. We found a significant spatial component, and our results suggested that the estimated range of influence and the overall location of areas with increased prevalence are not very sensitive to diagnostic misclassification.