Analytical goals for imprecision derived from data on intra-individual variation for total T4, free T4, total T3, free T3 and thyrotropin (TSH) are coefficients of variation (CV) less than or equal to 2.5, 4.7, 5.2, 3.9 and 8.1%, respectively. If total T4 is used to monitor replacement therapy with thyroxine, a more stringent goal of CV less than or equal to 1.4% is appropriate. For those analytes for which biological variation data are not available, analytical goals may be derived either from reference intervals or from the 'state of the art' as judged from the performance of a stated proportion of laboratories participating in an interlaboratory quality assessment scheme. Analytical goals for imprecision for reverse T3 and thyroxine-binding globulin derived from reference values are CV less than or equal to 10.7 and 7.2%, respectively. The goal for inaccuracy is that there should be none. Statements regarding the detection limit of an assay should be replaced with information about the range of concentrations over which specified goals for imprecision are met. If goals are not achieved at concentrations which are used for clinical decision making the 95% confidence limits of the extreme values of the 'working range' should be calculated using the relevant imprecision. Improved analytical performance will result in better between-laboratory comparability and eventually allow the use of universally applicable reference values.