For a sample of metal-poor stars （-3.3 ≤ [Fe/H] ≤-2.2） that have highresolution spectroscopic abundance determinations, we have measured equivalent widths of the Ca II K, Mg I b and near-infrared Ca II triplet lines using low-resolution spectra from the Sloan Digital Sky Survey （SDSS）, calculated effective temperatures from （g - z）o color, deduced stellar surface gravities by fitting stellar isochrones, and determined metallicities based on the aforementioned quantities. Metallicities thus derived from the Ca II K line are in much better agreement with the results deter- mined from high-resolution spectra than the values given in the SDSS Data Release 7. The metallicities derived from the Mg I b lines have a large dispersion owing to the large measurement errors, whereas those deduced from the Ca II triplet lines are too high due to both non-local thermodynamical equilibrium （NLTE） effects and mea- surement errors. Abundances after correction for the NLTE effect for the Mg I b lines and Ca II triplet lines are also presented. Following this method, we have identified six candidates of ultra-metal-poor stars with [Fe/H] - -4.0 from a sample of 166 metal-poor star candidates. One of them, SDSS J102915＋172927, was recently con- firmed to be an ultra-metal-poor （[Fe/H] 〈 -4.0） star with the lowest metallicity ever measured. Follow-up high-resolution spectroscopy for the other five ultra-metal-poor stars in our sample will therefore be of great interest.
