Contaminated soils can become exposure routes of elements toxic to human beings. The health risk of a toxic element by ingestion depends on its bioavailability in the gastrointestinal system, measured in vivo or in vitro. This study aimed to use a novel, versatile reactor (gastrointestinal simulation reactor system to determine bioaccessibility -GSRSB-) to measure lead and arsenic bioaccessibility in the gastric and intestinal phases by applying a modified physiologically based extraction test (PBET). Three composite samples of polluted soils with As (0.50 - 3.25%) and Pb (0.02 - 0.10%) and the certified reference material NIST 2710 were analyzed with this GSRSB-PBET method and the NOM-147 Mexican standard method, which uses an end-over-end shaker. All results were compared to one another. The NIST 2710 results were contrasted with those reported in vivo and in vitro by 14 laboratories. The (GSRSB-PBET) gastric phase ranges were 35.9-55.1 % (As) and 59.6-96.1 % (Pb), while (NOM-147) gastric phases were 35.8-60.4 % (As) and 61.0-70.7 % (Pb). The (GSRSB-PBET) intestinal phase ranges were 39.5-46.9 % (As) and 19.9-31.5 % (Pb). The As and Pb compounds and the stirring technique seem to influence bioaccessibility. On the other hand, the comparison of NIST 2710 results with those reported in vitro and in vivo indicated that As and Pb gastric bioaccessibility obtained with GSRSB-PBET falls into the in vivo results range, while NOM-147 results are higher and fall outside the in vivo range, possibly overestimating the risk. Thus, the proposed method is adequate for modifying the current Mexican Standard Method (NOM-147), which only allows the calculation of Pb gastric bioavailability in vitro.