Résumé : The prevalence of Anti-Neutrophil Cytoplasmic Antibodies (ANCA) directed against myeloperoxidase (MPO) in pauci-immune necrotizing crescentic glomerulonephritis (NCGN) is dependent on the assay(s) used. We investigated the frequency of MPO-ANCA as detected by different assays for MPO-ANCA in a large cohort of patients with biopsy-proven pauci-immune NCGN. Sera from 121 consecutive untreated patients presenting with pauci-immune NCGN were tested for ANCA directed to proteinase-3 (PR3) at diagnosis. PR3-ANCA negative sera were tested by direct ELISA using recombinant or native MPO and by capture ELISA using two different specific monoclonal antibodies directed to MPO and three different antigenic sources. Sera from 80 relevant disease controls were tested to explore the specificity of the different assays. Thirty-eight out of 121 patients (31%) with pauci-immune NCGN did not have PR3-ANCA. Sufficient amounts of serum from 30 of these 38 PR3-ANCA negative patients were available for further testing. Recombinant and native MPO were recognized by similar numbers of sera in a direct ELISA (recombinant MPO: 93%, native MPO: 93%) and a capture ELISA (recombinant MPO: 77-87%, native MPO: 93%). Sera of patients with PR3-ANCA positive pauci-immune NCGN and disease controls were less frequently positive for MPO-ANCA in a capture ELISA (recombinant MPO: 3-7%, native MPO: 6-7%) than in a direct ELISA (recombinant MPO: 25%, native MPO: 13%). Both direct and capture ELISA assays using either native or recombinant MPO are sensitive techniques to detect MPO-ANCA in patients with pauci-immune NCGN. A capture ELISA performs better than a direct ELISA because it combines a higher specificity with a comparable sensitivity. Recombinant MPO is a good alternative for native MPO when used as antigen in a capture ELISA, but not when used in a direct ELISA because of lower specificity in this latter assay. © 2001 Elsevier Science B.V.