Immunoprecipitation (IP) of radiolabeled PMN extracts was used as the gold standard for anti-proteinase 3 (PR-3) autoantibody detection to validate immunofluorescence (IF) and alpha granule (alpha) ELISA. A myeloperoxidase (MPO) ELISA was also used in parallel. We studied 48 patients with strictly defined vasculitic syndromes in the initial active phase of their disease. The 3 methods confirmed the high (> 90%) sensitivity and specificity of anti-PR-3 for patients with Wegener's granulomatosis (WG). Similarly, a high (86%) sensitivity of MPO-ELISA was found in microscopic polyarteritis as defined. Using alpha-ELISA, we could not improve the detection rate of anti-PR-3 obtained by IF-cANCA-pattern reading. Moreover, a small proportion (< 15%) of biopsy-proven WG patients had anti-MPO antibodies detected by IF, usually as pANCA but also, even if rarely, as bona fide cANCA (< 5%). Thus, IF would seem to be the most reliable screening method and alpha-ELISA should be used for confirmation. On the other hand, because MPO-ELISA detected twice as many anti-MPO positive sera as did pANCA pattern reading by IF, we suggest that in the clinical context of a vasculitis, MPO-ELISA should also be used as a screening test. Although IP is not designed for routine clinical use, it should be required when reporting the presence of anti-PR-3 in vasculitis-like diseases that are fertile grounds for false positive reactions.