We have assessed a new method of free T4 measurement (Amerlex) which uses a novel unidentified T4-labelled analogue, said to be unreactive with T4 binding proteins in serum, together with an antibody that binds both analogue and T4. Free T4 is assessed by competition with analogue for antibody binding-sites. The test method has been compared with free T4 measured by equilibrium dialysis and with a technique using an immobilized T4 antibody. All methods gave the expected free T4 levels in normal, hyperthyroid and hypothyroid subjects and normal free T4 levels with high or low levels of T4 binding globulin. However, in autosomal dominant familial euthyroid T4-excess, where T4 is abnormally bound to albumin, the test method gave apparent high free T4 levels suggestive of hyperthyroidism. In a selected group of severely-ill euthyroid patients the new method gave apparent low free T4 levels. In view of these discrepancies, binding of labelled analogue was evaluated by dextran-charcoal separation of 4 degraees C. Familial euthyroid T4-excess sera showed greater analogue binding and samples with low prealbumin concentration showed less binding than did normal sera. Despite its validity with variations in TBG, it appears that Amerlex Free T4 is influenced by lower-affinity, high-capacity T4 binding sites in serum, so that apparent free T4 concentration may vary with changes in the concentration of such sites.