A mouse assay for measuring the relative bioavailability (RBA) of arsenic (As) in soil was developed. In this study, results are presented of RBA assays of 16 soils, including multiple assays of the same soils, which provide a quantitative assessment of reproducibility of mouse assay results, as well as a comparison of results from the mouse assay with results from a swine and monkey assay applied to the same test soils. The mouse assay is highly reproducible; three repeated assays on the same soils yielded RBA estimates that ranged from 1 to 3% of the group mean. The mouse, monkey, and swine models yielded similar results for some, but not all, test materials. RBA estimates for identical soils (nine test soils and three standard reference materials [SRM]) assayed in mice and swine were significantly correlated (r = 0.70). Swine RBA estimates for 6 of the 12 test materials were higher than those from the mouse assay. RBA estimates for three standard reference materials (SRM) were not statistically different (mouse/swine ratio ranged from 0.86-1). When four test soils from the same orchard were assessed in the mouse, monkey, and swine assays, the mean soil As RBA were not statistically different. Mouse and swine models predicted similar steady state urinary excretion fractions (UEF) for As of 62 and 74%, respectively, during repeated ingestion doses of sodium arsenate, the water-soluble As form used as the reference in the calculation of RBA. In the mouse assay, the UEF for water soluble As(V) (sodium arsenate) and As(III) (sodium [meta] arsenite) were 62% and 66%, respectively, suggesting similar absolute bioavailabilities for the two As species. The mouse assay can serve as a highly cost-effective alternative or supplement to monkey and swine assays for improving As risk assessments by providing site-specific assessments of RBA of As in soils.