By immunoblot techniques, detailed antibody studies were performed with sera of 20 honeybee-venom-allergic patients during or at the end of specific immunotherapy (median duration: 3 years) and before honeybee sting challenge. Before immunotherapy, all patients had experienced systemic allergic reactions to a honeybee sting, with a mean severity of 3.5 +/- 0.5 according to the Müller classification. After the sting challenge, 10 patients (reactors) reacted again with a systemic allergic reaction, whereas 10 patients (nonreactors) did not. No differences were observed between reactors and nonreactors in total serum IgE and specific IgE to honeybee venom at the time of challenge. For immunoblot, honeybee venom (RELESS) was separated on 7.5-20% SDS-PAGE. For detection of specific IgE, IgG, IgG1, IgG4, and IgM, an alkaline phosphatase-linked second antibody was used. Both groups showed 11 antibody-binding bands: at 52, 46, 40, 31, 18.7, 16.9, 13, 11, 10, 9, and 8 kDa; however, the antibody-binding pattern was individual. The reactors differed from nonreactors in showing intense IgE and less IgG4 binding to at least one single component of the venom extract. For nonreactors, the inverse relationship was observed. The hypothesis, "intensity of IgE > or = IgG4 leads to allergic symptoms", was highly significant (P = 0.00026; chi-square). These immunoblot findings could offer predictive value in distinguishing reactors from nonreactors.