Plasmas from 40 haemophilia A patients enrolled in a study by the paediatric group of the German Society on Thrombosis and Hemostasis were tested by the Bethesda assay for inhibitor antibodies and by a more sensitive immunoprecipitation assay (IP) for all antifactor VIII antibodies. Of the 26 severe, 11 moderate and three mild haemophiliacs, 18, two, and none, respectively, had positive Bethesda titres after several factor VIII infusions. In 275 plasmas with Bethesda titres of 0, 0.6--1.0, > 1--5, and > 5--655, the IP responses were 0-238, 0--61, 0--786, and 43--6141, respectively, and a reliable positive IP titre was > 4.2. The overlapping ranges of IP titres indicated large differences in the ratio of inhibitory to noninhibitory antibodies in individual plasmas. In five of seven patients with Bethesda titres of 0.6--1, the IP titres were < 4.2, suggesting a lack of precision of Bethesda titres < or = 1. Detection of the primary immune response was found in only three patients by IP assay before a positive Bethesda assay. This precludes early, reliable testing of which patients will be immunologically responsive. In four patients undergoing immune tolerance therapy, antifactor VIII antibodies were still detectable by the IP assay in the absence of a Bethesda titre, which indicates that antibodies were completely eradicated in none of the patients. Our results show that the use of both the Bethesda and IP assays can provide more accurate detection of antifactor VIII antibodies in all patients.