A number of studies have found that gastrointestinal absorption of arsenic from soil is limited, indicating that a relative oral bioavailability (RBA) adjustment is warranted when calculating risks from exposure to arsenic-contaminated soil. However, few studies of arsenic bioavailability from soil have been conducted in animal models with phylogenetic similarity to humans, such as nonhuman primates. We report here the results of a study in which the RBA of arsenic in soil from a variety of types of contaminated sites was measured in male cynomolgus monkeys. A single oral dose of each contaminated soil was administered to five adult male cynomolgus monkeys by gavage, and the extent of oral absorption was evaluated through measurement of arsenic recovery in urine and feces. Urinary recovery of arsenic following doses of contaminated soil was compared with urinary recovery following oral administration of sodium arsenate in water in order to determine the RBA of each soil. RBA of arsenic in 14 soil samples from 12 different sites ranged from 0.05 to 0.31 (5-31%), with most RBA values in the 0.1-0.2 (10-20%) range. The RBA values were found to be inversely related to the amount of arsenic present with iron sulfate. No other significant correlations were observed between RBA and arsenic mineralogic phases in the test soils. The lack of clear relationships between arsenic mineralogy and RBA measured in vivo suggests that gastrointestinal absorption of arsenic from soil may be more complex than originally thought, and subject to factors other than simple dissolution behavior.