Investigating the humoral immune response to mucosal antigens in patients with lung cancer, we have documented a preferential immunoglobulin G (IgG) binding to cryptic epitopes unmasked by the proteolysis of bovine beta-lactoglobulin (BLG). In contrast, IgG from healthy controls and patients with chronic bronchitis (COPD) bind preferentially to continuous epitopes presented on both native (n) and denaturated (d) forms of this antigen. The present study further characterized the differences in the epitope profiles recognized on BLG. The capacity of individual sera from 65 lung cancer patients, tested before and after cancer removal for the patients with early stage lung carcinoma, 65 healthy controls, and 52 patients with COPD, to prevent the binding of pooled IgG fractions from each population as well as murine monoclonal antibodies (MoAb), specific for BLG, to solid phase bound antigen was evaluated in enzyme-linked immunoadsorbent assay using streptavidin-biotin technology. Some of these experiments were also performed with sera from 42 patients diagnosed with other cancers. Compared with control sera and sera from patients with other solid tumors, lung cancer patient sera showed distinct capacities to prevent the binding of murine MoAb as well as human pooled IgG fractions to n- and d-BLG. The inhibition capacities of lung cancer sera changed as soon as five weeks after cancer removal. The results indicate that the difference in epitope specificity exhibited by lung cancer sera is not restricted to cryptic epitopes, but also affects continuous and discontinuous epitopes, accessible only on the native antigen. A high level of binding discrimination between antibodies from the study populations is also observed at the level of the epitope. This deviation in the epitope specificity of antibodies changes soon after cancer removal, suggesting a tumor-dependent disturbance. Also documented in the Dermatophagoides pteronyssinus model, it opens the way to a new class of paraneoplastic immune markers for this malignancy, with, at first glance, a high specificity level.