The association of IgA anti-gliadin antibodies and IgA glomerulonephritis (IgA GN) was first reported in 1987 (Am J Nephrol, 1987, 7, 178-183) and has since been confirmed by other groups. We have developed a second generation ELISA (alkaline phosphatase, biotin-avidin) and used it to test 45 adult IgA GN, 34 idiopathic membranous nephropathy (MN), 31 idiopathic nephrotic syndrome (INS), and 11 idiopathic membranoproliferative glomerulonephritis (MPG) patients. IgA anti-gliadin antibodies were found in 24 IgA GN (53%), 1 MN (3%), 1 INS (3%), and 1 MGP (9%) patients. The presence of these antibodies in a patient with proteinuria strongly suggests IgA GN, with a sensitivity of 53%, a specificity of 96%, a positive predictive value of 88% and a negative predictive value of 77%. The presence of IgA anti-gliadin antibodies in IgA GN did not necessarily indicate coeliac disease because: a) neither IgG nor IgA anti-reticulin nor IgA anti-endomysium antibodies were found; b) intestinal absorption tests (folates, EDTA) were normal; c) biopsies of the small intestine were normal; and d) a gluten-free diet did not alter the evolution of the disease. Immunochemical analysis (footprinting after separation of the gliadins by rocket electrophoresis) showed the variability of the fractions recognized by the IgA antibodies from patients and controls, in addition to the absence of a typical profile. Gliadin does not have a lectin effect, since mannan and mannose did not inhibit the ELISA. Immunofluorescent labeling of human kidney with purified rabbit IgG anti-gliadin antibodies did not reveal a common epitope shared by gliadin and renal structures.(ABSTRACT TRUNCATED AT 250 WORDS)