Three sample introduction systems were compared for their efficiency in introducing a variety of biologically relevant selenium species in organic solutions with the purpose of mimicking the effluent from a chromatographic system. The sample introduction systems consisted of a microconcentric nebuliser (MCN) in combination with a membrane desolvator ( MD), a MCN in combination with a cyclonic spray chamber ( CS) and a direct injection nebuliser ( DIN), respectively. Thirteen selenium standards diluted in 50% methanol were introduced in each system at a flow rate of 50 mu l min(-1). Except for the selenite and methaneseleninic acid standards that were lost in the MCN-MD system, similar normalized response factors were achieved for the standards regardless of the sample introduction system. When trimethylselenonium and selenate standards were continuously infused in a linear methanol gradient from 10 to 90%, the signal was almost unaffected when the MCN-MD was used as interface. The signals from the MCN-CS introduction system were almost unaffected in methanol concentrations between 10 and 40%, after which they slowly declined. The signal from the DIN showed a plateau in signal intensity at methanol concentrations between 35 and 70%. Detection limits in flow injection analysis, expressed as the concentration that would result in a signal equivalent to three times the peak to peak noise of the baseline, were estimated to be 200, 150 and 100 ng Se L-1 for the MCN-CS, MCN-MD and DIN, respectively. The independence of the signal from the methanol concentration in the eluent allowed quanti. cation of human selenoprotein P isoforms by gradient microbore reversed phase chromatography (mu RPC) on the basis of an internal trimethylselenonium standard. The MCN-MD system was applied for mu RPC analysis of Se-77 enriched yeast samples and more than 30 selenium containing compounds were separated using a linear gradient from 5% to 95% methanol, some of the selenium species eluting in 85% methanol