Excessive bias and imprecision are major analytical problems associate
d with some assays for free and total thyroxine (T-4). Bias in free T-
4 methods is largely proportional to variations in serum T-4 binding.
In direct methods, this is attributable to requirements for substantia
l quantities of protein-bound T-4 to replace analytical losses of free
T-4. In some total T-4 methods, bias is inversely proportional to the
amount of serum T-4 binding and is attributable to the incomplete rel
ease of serum protein-bound T-4. In others, bias is fixed and attribut
able to inaccurate calibration. Manufacturers should report the bias i
n their methods. Calibrations should be standardized, Imprecision vari
es widely among methods, but is generally less for total T-4 methods t
han for free T-4 methods. A consensus on quantitative analytical perfo
rmance goals for free and total T-4 methods mould be helpful. Here, pe
rformance goals are proposed, based in part on the best achievements o
f current methods.