The need for a routine, inexpensive method with nanomolar detection limits for soluble SeO32- and SeO42- is increasing as Se research on soil-plant-water systems intensifies and speciation of Se valence states is required. This study was undertaken to compare hydride generation-atomic absorption spectrometry (HGAAS), colorimetry using 2,3-diaminonaphthalene (DAN), and ion chromatography (IC) as suitable methods for analysis of SeO32- and SeO42- in soil extracts and water samples. Selenium standards in solutions of either 1.0 mM KH2PO4, CaSO4, and KNO3, or in distilled water were analyzed to determine detection limits (DL), limits of quantitation (LOQ), and practicality for routine use of the above methods. A 0.1 mM K2HPO4/KH2PO4 soil extract spiked with 2.0 or 20 muM SeO32- and SeO42- was also used to compare the precision of the three methods. The HGAAS method was the most sensitive (0.212 absorbance units/muM) and had the lowest DL for SeO32- (0.03 mumol Se/L) of the three methods employed. Detection limits for the DAN and IC methods were significantly higher (18 and 0.24 mumol Se/L, respectively). Poor reproducibility between methods was obtained for SeO42- content of the soil extract containing 20 muM SeO42-, with concentrations ranging from 9.1 +/- 1.2 mumol/L determined by HGAAS to 24.6 +/- 0.05 mumol/L by IC. The variability and poor recovery of SeO42- was due to variabilty in efficiency of the HCI reduction required to reduce SeO42- to SeO32- before analysis in the HGAAS and DAN procedures. Selenium determinations in samples where low detection limits (<0.1 muM) are required are most effectively accomplished with HGAAS. Colorimetry requires greater effort and time from the analyst to obtain the low detection limits of HGAAS, but is useful in that only a visible light spectrometer is required. Ion chromatography is capable of automated determinations of SeO32- and SeO42- directly in combination with other ions of interest and may be appropriate for soil solution Se analysis where Se speciation is required.