Recent work has shown the presence of volatile selenium metabolites in human urine and suggested that these compounds could compromise quantitative selenium analyses by ICPMS. We show that with a commonly used sample introduction system (pneumatic nebulizer and spray chamber), two volatile selenium species recently identified in urine, namely, dimethyl selenide and dimethyl diselenide, gave greatly increased ICPMS responses (up to 58-fold) relative to selenite, an effect related to their volatilization in the spray chamber resulting in enhanced transport to the plasma. The quantitative consequences of this effect were demonstrated by measurement of total selenium and selenium species in certified reference material, NIES CRM 18 human urine. Direct flow injection analysis of the urine gave a total selenium concentration more than 2-fold higher than the certified value. These data suggested that NIES CRM 18 may contain part of its selenium as volatile species, and subsequent reversed-phase HPLC/ICPMS showed the presence of dimethyl selenide in addition to selenosugars and trimethylselenonium ion. Although the practice of quantifying unidentified chromatographic peaks against those of known compounds is common in speciation analysis, this approach when applied to NIES CRM 18 gave a value for the sum of selenium species which was twice the certified total selenium concentration. This work shows that the presence of volatile selenium species in urine precludes the use of flow injection analysis for total selenium measurements and imposes severe restrictions on the quantification of urinary selenium metabolites. In addition, it raises broader issues of the validity of the "dilute and shoot" approach to the determination of metals in clinical analysis of biological fluids.