Measurement of thyroglobulin (Tg) has become an established part of the monitoring of patients with differentiated thyroid cancer. Numerous studies have shown that Tg results that are very low or undetectable with modern assays (typically with detection limits of around 0.1–0.2 ng/ml), either at baseline or after stimulation with recombinant TSH, indicate individuals with a small burden of disease and a low likelihood of progression. Tg is measured in almost all laboratories using noncompetitive, immunometric assays; these are preferred because they are easily automated, are readily available, and can detect lower amounts of Tg than older, competitive immunoassays (which are not easily obtained from manufacturers of testing kits). A major limitation in the use of Tg is interference from Tg antibodies. In immunometric assays, Tg antibodies can lead to falsely low results; current guidelines recommend routine measurement of Tg antibodies at the time of Tg measurement and recommend caution in the interpretation of results when antibodies are present (1). Although about 10% of the general population has Tg antibodies, they are more frequently found in patients with thyroid cancer. Tg antibodies have been found to be prognostic in persons with thyroid cancer, with falling antibody levels associated with a low risk of recurrent disease, whereas rising or continually positive antibodies indicate increased risk for recurrent cancer. They have also been found to predict risk of cancer in thyroid nodules (2). The study by Spencer et al. (3) in this issue of JCEM casts serious doubt on the reliability of assays to detect Tg antibodies in patients with thyroid cancer. More importantly, their study also shows that the assays fail to detect antibodies that cause significant underestimation of Tg results in many patients. Because Tg antibodies typically do not cause falsely low results in serum Tg concentrations when measured by the research competitive immunoassay used in the University of Southern California laboratory, samples with Tg levels at least 25% lower using commercial assays than with the research assay were considered to show interference from Tg antibodies. Using the cutoffs suggested by the manufacturers for four commercial Tg antibody assays, they showed that between 21 and 44% of specimens termed antibody negative had falsely low Tg, and between 4 and 10% of samples had undetectable Tg when results were clearly detectable by the research assay. Why do Tg antibody assays fail to recognize interfering antibodies? One reason, as outlined by Spencer et al. (3), is that the cutoff limit is set too high to detect interfering antibodies. Cutoff limits for laboratory tests are typically determined as the level at which only 2.5% of “normal” people will have an increased result. For Tg antibodies, this is based on excluding persons with autoimmune thyroid disease from being classified as “normal,“ although data from the National Health and Nutrition Examination Survey III study suggest that Tg antibodies detected using this cutoff are not correlated with risk of thyroid dysfunction (4). In the study of Spencer et al. (3), many of the samples that were classified as antibody negative did, in fact, have measurable Tg antibodies (but were below the cutoff value determined by the manufacturer); 50–80% of these showed interference in Tg measurement, and up to 20% of these had undetectable Tg. For two of the assays, changing the cutoff limit to the lowest amount that could be measured virtually eliminated samples that showed interference in Tg measurement; however, for the other two assays, samples with interference were still not detected using this value, and 7–12% had undetectable Tg (but positive results by the research assay). Another factor is the complexity of the Tg molecule. Tg has a molecular mass of around 660 kDa and has many