This paper summarises the results of the evaluation of several assays of thyrotropin in serum. The authors assessed 8 radioimmunoassays (IRMA techniques) and 10 enzyme immunoassays, including ELISA, luminescent immunometry and fluoroimmunometry, also covering a broad spectrum of actual assays. An unique protocol based on identical human samples for all assays was used in order to compare precision, linearity, detection limit, matrix effects and correlation. The authors found a wide spread in precision results between the evaluated assays; as distinct from enzyme immunoassays, discrepant analytical results of the radioimmunoassays are mainly due to differences in standardization procedures. All assays were linear in a smaller or larger range. Sensitivity was generally in agreement with the manufacturer's specification. One assay was not acceptable for clinical use because the practical detection limit only reached 1.5 mU/l. Three other assays (from the same origin) demonstrated particular problems with the stability of the chromogenic reagent. The study of matrix effects was performed using different artificial matrices and the results varied from one assay to another; in the test conditions, some assays were not sensible to matrix manipulations, others were extremely influenced by ion strength or protein concentration. The correlation study confirmed the results of the precision study: assays with a good intraassay precision had also the best correlations. In a general appreciation of the evaluated methods the authors notice that most of the evaluated "classical" enzyme immunoassays operate around the sensitivity limits of spectrophotometry, explaining also the inferior results. Some radioimmunoassays and the enhanced luminescent technique gave the best results in this evaluation study