Soil ingestion by children is an important pathway in assessing public health risks associated with exposure to arsenic-contaminated soils. Soil chemical methods are available to extract various pools of soil arsenic, but their ability to measure bioavailable arsenic from soil ingestion is unknown. Arsenic extracted by five commonly used soil extractants was compared with bioavailable arsenic measured in vivo by immature swine (Sus scrofa) dosing trials. Fifteen contaminated soils that contained 233 to 17 500 mg kg(-1) arsenic were studied. Soil extractants were selected to dissolve surficially adsorbed and/or readily soluble arsenic (water, 1 M sodium acetate, 0.1 M Na2HPO4/0.1 M NaH2PO4) and arsenic in Fe and Mn oxide minerals (hydroxylamine hydrochloride, ammonium oxalate). The mean percent of total arsenic extracted was: ammonium oxalate (53.6%) > or = hydroxylamine hydrochloride (51.7%) > phosphate (10.5%), acetate (7.16%) > water (0.15%). The strongest relationship between arsenic determined by soil chemical extraction and in vivo bioavailable arsenic was found for hydroxylamine hydrochloride extractant (r = 0.88, significant at the 0.01 probability level). Comparison of the amount of arsenic extracted by soil methods with bioavailable arsenic showed the following trend: ammonium oxalate, hydroxylamine hydrochloride > in vivo > phosphate, acetate > water. The amount of arsenic dissolved in the stomach (potentially bioavailable) is between surficially adsorbed (extracted by phosphate or acetate) and surficially adsorbed + nonsurficial forms in Fe and Mn oxides (extracted by hydroxylamine hydrochloride or ammonium oxalate). Soil extraction methods that dissolve some of the amorphous Fe, such as hydroxylamine hydrochloride, can be designed to provide closer estimates of bioavailable arsenic.