The determination of thyroid hormones is widely used for the diagnosis and therapy control of thyroid disorders. In particular, thyroxine in serum is one of the most frequently determined endocrine parameters. Unfortunately, the results obtained by the use of different commercial test kits vary significantly, and until now there has been no means to decide whether a particular enzyme or radioimmunoassay kit yields accurate results or not. It seemed, therefore, necessary to develop definitive or reference methods for the measurement of thyroxine and to apply this technique for the assessment of target values in control sera for internal and external quality control. In the present investigation, an analytical protocol using the isotope dilution mass spectrometry technique is described which is herewith proposed as a definitive method for the measurement of thyroxine in human serum. The procedure consists of the following steps. (i) Equilibration of endogenous thyroxine in a serum sample with 100 ng ["C,]thyroxine. (ii) Isolation of the thyroid hormones by using a cation exchange resin. (iii) Formation of the methyl ester by reaction with methanolic hydrochloric acid. (iv) Purification of the methyl ester by column chromatography on Sephadex LH-20. (v) Formation of the N,O-bistrifluoroacetyl derivative with trifluoracetic anhydride. (vi) Selected ion monitoring of fragment ions of the thyroxine and the ['3C21thyroxine derivatives a t m/z 870 and 872 using a magnetic sector field mass spectrometer with electron impact ionization combined with a capillary gas chromatography column. This method is now used to assign target values in a German quality control scheme. The precision of the method is of the order of 1-2% (coefficient of variation).