Intake of contaminated soils is considered as an important exposure pathway of polybrominated diphenyl ethers (PBDEs) to humans, especially for children during their outdoor hand-to-mouth activities. Oral bioaccessibility is an essential tool to quantitatively assess the exposure risk of pollutants. In this study, we employed an in vitro digestion model to mimic the gastrointestinal digestion of typical PBDEs (BDE-28, BDE-47, BDE-99 and BDE-153 at a series of initial concentrations) in three natural soil samples with different TOC contents and to verify a previous hypothesis that the sorption of PBDE fraction mobilized from soil into digestive fluid on the surface of residual solid phase may lead to an underestimation of bioaccessibility of PBDEs. In addition, a method with multiple fluid-to-solid ratios was applied to calibrate the underestimation. The results indicated that the calibrated digestibility values were commonly higher than those without correction. For the different soil samples, the averaged increasing rates of PBDE digestibility at different initial concentrations ranged from 14.3% to 42.3%, from 11.1% to 32.1%, from 4.9% to 12.3% and from 0.0% to 7.7% for BDE-28, BDE-47, BDE-99 and BDE-153, respectively. Therefore, the bioaccessibility of PBDEs in gastrointestinal gut would be significantly underestimated without calibration, especially for tri- and tetra-BDEs and soil samples with low TOC contents or high PBDEs concentrations. The corrected digestibility values of BDE-28, BDE-47, BDE-99 and BDE-153 were 21.9%-54.7%, 18.8%-43.1%, 13.4%-27.2% and 9.3%-19.9%, respectively. The results indicated that the PBDEs digestibility was negatively correlated with lgK(ow); whereas there was no significant correlation of PBDE bioaccessibility with TOC contents in soils or with initial concentrations of PBDEs, particularly for the highly brominated components.