Thyroid disorders can be associated with elevated concentrations of serum antithyroglobulin antibodies (anti-Tg) and/or thyroglobulin (Tg), but none of the available anti-Tg assays deals with anti-Tg measurements in the presence of abnormally high Tg levels. The competitive binding radioassay produces falsely elevated values for anti-Tg if serum Tg is elevated and either falsely elevated or depressed values if both Tg and anti-Tg are abnormally high. The falsely elevated anti-Tg values can be identified by measuring the formation of Tg[125I]anti-Tg complexes in the supernatant of the anti-Tg assay (supernatant assay). For screening purposes, we modified the original anti-Tg RIA into a solid phase, sandwich-type RIA. Anti-Tg in serum or standard is first bound to plastic cups coated with Tg and then quantitated by binding of [125I]Tg. This assay has a detection limit of 2 U/ml serum, intra- and interassay coefficients of variation of 9--15%, and a normal range of less than 2 U/ml. When sera with normal Tg concentrations were analyzed, the results for anti-Tg obtained by the competitive binding RIA and the new sandwich RIA were comparable as far as positives and negatives were concerned, and the numerical values for positive sera correlated moderately well (r = 0.79); the sandwich assay, in general, gave lower values for anti-Tg. The major advantages of the sandwich anti-Tg RIA are the elimination of false positive results and its applicability to sera containing high levels of both Tg and anti-Tg. In the latter case, the results indicate the level of free anti-Tg present, as opposed to antibody present in the form of Tg-anti-Tg complexes.