This study was designed to investigate quantitatively the interference of thyroglobulin autoantibodies in the RIA of human thyroglobulin (hTG). Anti-hTG autoantibodies were combined with purified hTG to produce samples with known antibody titers and hTG concentrations. These samples were analyzed in the RIA. By using anti-human globulin serum it was first shown that immune complexes formed between labeled hTG and human anti-hTG. It was then shown that the most important factor in determining the direction of the interference was the specificity of the precipitating (second) antiserum with respect to these immune complexes. When the precipitating antiserum was specific, i.e. did not recognize human antibodies, the immune complexes remained in the supernatant and the measured hTG concentration was falsely elevated. When the precipitating antiserum cross-reacted with human antibodies, the direction of the interference depended on the sample volume. At small volumes there was false depression while at large volumes there was false elevation of apparent hTG levels, depending on the capacity of the precipitating antiserum to combine with human antibodies. Anti-hTG titers far below those detected by the tanned-red cell hemagglutination test had very large effects, to the point where measurements of hTG could not be made, when a cross-reactive precipitating antiserum was used. Therefore, the procedure which investigators have used until now, to exclude samples with anti-hTG hemagglutination titers above an arbitrary limit, is not adequate. It is necessary, until methods are developed which avoid the problem of autoantibody interference, to characterize each assay to determine the limits of anti-hTG that can be tolerated. The factors which influence anti-hTG interference in the hTG RIA are 1) the specificity of the precipitating antiserum, 2) the sample volume, 3) the maximum tracer binding, and 4) the anti-hTG titer.