In the absence of a "soil model" on arsenic bioavailability, many investigators conducting baseline risk assessments of arsenic-enriched sites have assumed that all arsenic present in the soil is bioavailable. This assumption overestimates actual human health risk because various geochemical forms of arsenic are stable, or insoluble, in human digestive juices. The authors conducted a laboratory incubation study to analyze the in vitro bioavailability of arsenic in soils as a function of soil properties. Four different soil types were selected on the basis of their potential differences with respect to arsenic reactivity. Each soil was amended with sodium arsenite at a rate representative of a routine 1-yr application of arsenical pesticide in an agricultural system. The soils were incubated for 1 yr, after which the authors measured soil-specific total and bioavailable arsenic concentrations. Results demonstrated that soil physicochemical properties significantly affect arsenic bioavailability, and hence estimates of cancer risk, which in turn affect site cleanup cost projections.