Published reports on selenium concentrations in serum and plasma were critically reviewed according to the criteria set up by the expert group of the international project TRACY. The aim was to examine both pre-analytical (sampling conditions and characteristics of the group) and analytical factors described in papers to establish reference values for evaluation of persons with deficiency or excessive exposure to the element. Out of 291 papers published during the period 1983 to 1993, which we reviewed, 36 were considered suitable for the TRACY project. In order to eliminate sources of variation due to age and health status, only data concerning healthy adults were retained. Of the sampling and other pre-analytical factors, geographical area and time of sampling were found to be of high importance. Gender, occupation, diet data and use of supplements or medications were also considered important determinants, but ethnic origin, smoking and fasting, contamination control and storage of sample were held to be of relatively low importance. Concerning analytical factors, accuracy criteria were generally satisfied although not extensively documented. Reference materials for serum selenium, which contribute to better analytical reliability have been available for over 10 years. Due to the geographical variation of selenium intake, one universal reference serum selenium value for the world's population was not feasible but the data selected allowed documentation of the serum selenium concentration levels among populations of different areas, mostly in Europe. Critical use of the published TRACY criteria in future papers will certainly extend their reliability and scientific significance.