This study was carried out to examine the optimal analytical method for determination of urinary toxic arsenic (inorganic arsenic and its metabolites) by HG-AAS (hydride generation-atomic absorption spectrometry). In the analysis of SRMs (standard reference materials), method E (addition of 0.4% L-cysteine to pre-reductant and use 0.04M HCl as carrier acid) showed the most accurate results compared with the reference values. In the analysis of 30 urinary samples, analytical results were significantly different depend on the component of pre-reductant and the concentration of carrier acid. When the concentration of carrier acid was higher, the analytical result was lower. The recovery rates of MMA (monomethylarsonic acid) and DMA (dimethylarsenic acid) were varied by the concentration of pre-treatment acid and carrier acid and hydride generation reagents. When the concentration of carrier acid was 1.62 M (5% HCl), the recovery rates of DMA was 1%. The recovery rates of MMA and DMA in method E (=V) were 102% and 100%, respectively. The results of this study suggest that the component and concentration of pre-reductant and carrier acid must be carefully adjusted in the analysis of urinary arsenic, and method E is recommendable as the most precise analytical method for determination of urinary toxic arsenic.