Two selenium metabolites, Se-methylseleno-N-acetylgalactosamine (SeGal-N-Ac) and Se-methyl-selenogalactosamine (SeGal-NH2), were quantified in human urine by LC-DRC-ICP-MS. Urine samples were analysed after 1+1 dilution in a reversed phase chromatographic system using an eluent consisting of 200 mM ammonium acetate and 5% methanol with a pH of 9.25 and quantified by standard addition. Samples were collected from 8 volunteers before and after 5 days ingestion of 100 mu g Se day(-1) in form of selenized yeast. The average concentration of (SeGal-NH2) before and after selenium intake was 1.4 and 1.9 mu g Se L-1, respectively, while the average concentration of Se-Gal-N-Ac increased from 2.6 to 11.6 mu g Se L-1 before and after selenium consumption. Detection limits calculated on basis of three times the standard deviation on peak areas of 2 mu g Se L-1 solutions were 0.1 mu g Se L-1 for SeGal-NH2 and 0.2 mu g Se L-1 for SeGal-N-Ac based on peak areas and monitoring Se-80. The precision expressed as the relative standard deviation (n=6) at the 2 mu g Se L-1 level was 3.1 and 1.7% for SeGal-NH2 and SeGal-N-Ac, respectively, while the corresponding values were 1.0 and 0.7% at the 10 mu g Se L-1 level. Linearity in urine matrix was examined in the range 0.5-100 mu g Se L-1 and correlation coefficients better than 0.999 were obtained. As the cationic compound SeGal-NH3+ may be confounded with the trimethylselenonium ion (TMSe+), urine samples were also analysed in a cation exchange chromatographic system in which SeGal-NH3+ was separated from the trimethylselenonium ion. None of the samples contained TMSe+ in detectable amounts. Three sample introduction systems were compared-a microconcentric nebuliser in combination with a cyclonic spray chamber (MCN), a direct injection nebuliser (DIN) and an ultrasonic nebuliser (USN). The MCN was most suitable for this purpose