Stabilization of arsenic-contaminated soils with ferrous sulfate has been reported in many studies, but there are few stabilization effects assessments simultaneously combined chemical extraction methods and in vitro methods, and further explored the corresponding alternative relationships. In this study, ferrous sulfate was added at Fesingle bondAs molar ratio of 0, 5, 10 and 20 to stabilize As in 10 As spiked soils. Stabilization effects were assessed by 6 chemical extraction methods (toxicity characteristic leaching procedures (TCLP), HCl, diethylenetriamine pentaacetic acid (DTPA), CaCl2, CH3COONH4, (NH4)2SO4), and 4 in vitro methods (physiologically based extraction test (PBET), in vitro gastrointestinal method (IVG), Solubility Bioaccessibility Research Consortium (SBRC) method, and the Unified Bioaccessibility Research Group of Europe method (UBM)). The results showed that the HCl method provides the most conservative assessment results in non-calcareous soils, and in alkaline calcareous soils, (NH4)2SO4 method provides a more conservative assessment. In vitro methods provided significantly higher As concentrations than chemical extraction methods. The components of the simulated digestion solution as well as the parameters may have contributed to this result. The small intestinal phase of PBET and SBRC method produced the highest and lowest ranges of As concentrations, and in the range of 127–462 mg/kg and 68–222 mg/kg when the Fesingle bondAs molar ratio was 5. So the small intestinal phase of PBET method may provide the most conservative assessment results, while the same phase of SBRC may underestimate the human health risks of As in stabilized soil by 51 %(at a Fesingle bondAs molar ratio of 5). Spearman correlation analysis indicated that the small intestinal phase of PBET method correlated best with HCl method (correlation coefficient: 0.71). This study provides ideas for the assessment of stabilization efforts to ensure that stabilization meets ecological needs while also being less harmful to humans.