Bovine dictyocaulosis is a pulmonary parasitic disease present in temperate countries, with potential important clinical and economic impacts. The Baermann technique is routinely used despite its low sensitivity in adult cows. Recently developed serological tests seem to offer better sensitivity, but validations of these tests in field conditions are few. We aimed to study two non-previously evaluated diagnosis methods of dictyocaulosis based on bronchoalveolar lavage sampling (BAL), which allows finding lungworm stages in the lungs as well as determination of eosinophilia. We compared them to the Baermann technique and serological tests. As no gold standard was available, we performed a Bayesian analysis by the simultaneous use of latent class and mixture models. The study was carried out during the 2015 pasture season on 60 adult cows originating from 11 herds with clinical signs of dictyocaulosis, and 10 apparently healthy cows originating from the teaching herd of VetAgro Sup, in France. Prevalence of infection was highly variable among herds with clinical signs (10–90%). Despite a maximal specificity (100%), the sensitivity of parasitological methods was low (7.4% for the Baermann sedimentation and 24.7% for the examination of BAL fluids). Better results were observed with serology (Se = 74.9%, Sp = 85.5%) with an optimal cut-off value estimated at 0.397 for the optical density ratio. Even better results were obtained with the count of eosinophil in BAL (Se = 89.4%, Sp = 85.2%) with an optimal cut-off value estimated at 4.77% for the eosinophil proportion. The BAL is a relevant diagnostic method of dictyocaulosis for practitioners due to the opportunity to perform two analyses (direct parasitic research and the eosinophil count) and to its good sensitivity and specificity.