We compared four methods to extract arsenic (As) from three different plants containing different As levels for As speciation with the goal of developing a more efficient method, i.e., As-hyperaccumulator Pteris vittata at 459-7714mgkg(-1), rice seedling at 53.4-574mgkg(-1), and tobacco leaf at 0.32-0.35mgkg(-1). The four methods included heating with dilute HNO3, and sonication with phosphate buffered solution, methanol/water, and ethanol/water, with As being analyzed using high-performance liquid chromatography coupled with inductively-coupled plasma mass spectrometry (HPLC-ICP-MS). Among the four methods, the ethanol/water method produced the most satisfactory extraction efficiency (~80% for the roots and >85% for the fronds) without changing As species based on P. vittata. The lower extraction efficiency from P. vittata roots was attributed to its dominance by arsenate (82%) while arsenite dominated in the fronds (89%). The ethanol/water method used sample:solution ratio of 1:200 (0.05g:10mL) with 50% ethanol and 2h sonication. Based on different extraction times (0.5-2h), ethanol concentrations (25-100%) and sample:solution ratios (1:50-1:300), the optimized ethanol/water method used less ethanol (25%) and time (0.5h for the fronds and 2h for the roots). Satisfactory extraction was also obtained for tobacco leaf (78-92%) and rice seedlings (~70%) using the optimized method, which was better than the other three methods. Based on satisfactory extraction efficiency with little change in As species during extraction from three plants containing different As levels, the optimized method has the potential to be used for As speciation in other plants.