With an indirect immunofluorescence technique 77% of 96 patients with type I allergy and 40% of 20 patients with intrinsic bronchial asthma showed positive reactions for IgG anti-IgG antibodies in serum. They were present partly in an aggregated state not directly detectable before treatment with dithiothreitol. The aggregates could be removed by precipitation with polyethylene glycol. The IgG anti-IgG in hyposensitized patients were directed against both F(ab')2 and Fc fragments of rabbit IgG. Thirty of the type I allergic patients were examined once during hyposensitization as well. Before treatment 87% had IgG anti-IgG (titres 9-72). After greater than or equal to 13 months of treatment 100% were positive (titres 36-288). Eight patients were also examined after hyposensitization had been discontinued for at least 12 months. The titres of IgG anti-IgG had then reverted to the levels obtained before hyposensitization. Of 116 controls matched for sex and age, 7% had IgG anti-IgG antibodies. It is suggested that the production of IgG anti-IgG may be stimulated by the presence of immune complexes and that purity, amount and/or combination of allergens administered during hyposensitization may influence the production of anti-IgG antibodies. Neither IgE anti-IgG nor antinuclear antibodies seem to be of particular significance in allergic patients.