Abstract: The unified bioaccessibility model (UBM) was used to measure Arsenic (As) bioaccessibility of 13 soils with different physio-chemical properties and concentrations from Hunan, Guangxi and Dalian. Furthermore, the implications of incorporating As bioaccessibility into risk assessment were evaluated. The results revealed that the bioaccessibility of As in the stomach was between 3.9%-49.5%, with the arithmetic mean being 19.6%, while in the intestine it was 1.2%-10.8%, with the arithmetic mean being 6.0%. The bioaccessibility in the stomach was 1.2-9.1 times that in the intestine. The most significant factor controlling the bioaccessibility of As in the stomach was w(TAs) in soil (R2=0.94, P<0.01, n=13), followed by w(TP) (R2=0.82, P<0.01, n=13) and w(TMn) (R2=0.79, P<0.01, n=13). In the intestine, the most significant factor was also w(TAs) (R2=0.83, P<0.01, n=13), followed by w(TP) (R2=0.80, P<0.01, n=13), As bioaccessible concentration in the stomach (R2=0.76, P<0.01, n=13), pH (R2=0.74, P<0.01, n=13), w(TMn) (R2=0.65, P<0.02, n=13) and w(TOM) (R2=0.59, P<0.04, n=13). A regression model based on w(TAs) and w(clay) in soil was tested, and able to predict As bioaccessibility in the stomach with R2, ME, RMSE and r_p² being 0.97,0.02,0.17 and 0.95, respectively. For As bioaccessibility in the intestine, a model constructed based only on w(TAs) could be used to predict its bioaccessibility precisely, with R2, ME, RMSE and r_p² being 0.90,-0.03, 0.26 and 0.80, respectively. The health risk calculated based on w(TAs) was 2.0-15.0 times and 7.3-81.0 times the values when bioaccessibility in the stomach and intestine was incorporated, indicating the conservative nature of the current assessment approach assuming the bioaccessibility of contaminants in soil to be 100%.