Cleanup goals for sites affected by inorganic contaminants often are established on the basis of risk assessments, and these assessments rely on the estimated oral toxicity of the substances of concern. These toxicity estimates typically are based on historical studies in which a soluble salt of the metal was dissolved in water or mixed in food and then ingested by an animal or human. However, these toxicity studies do not account for the characteristics of a metal in soil or the limitations that these characteristics place on enteric absorption of that metal. Therefore, a more accurate risk assessment must account for the bioavailability of the metal in site-specific soil, relative to the bioavailability of the metal in the form administered in the toxicity study (i.e., the relative bioavailability of the element in soil). Historically, relative bioavailability estimates for metals in soil have been based on in vivo studies in laboratory animals. Given the costs and time constraints associated with such studies, it is clear that a more efficient alternative is desirable. The most promising option involves the development and validation of in vitro extraction tests that are predictive of oral metals bioavailability from soil. Such tests would provide a rapid and inexpensive method for developing more accurate exposure estimates for use in human health risk assessments. This paper reviews the site-specific in vivo studies that have been conducted to estimate the relative bioavailability of arsenic and lead in soil, discusses the soil and mineralogical factors that influence the bioavailability of these elements, and reviews the research to date on the development of bioavailability-predictive extraction tests for metals in soil. Finally, this paper outlines an ongoing collaborative research project to formally validate an in vitro extraction test for use in estimating the oral bioavailability of arsenic and lead in soil.