An in vivo swine assay was utilised for the determination of arsenic (As) bioavailability in contaminated soils. Arsenic bioavailability was assessed using pharmacokinetic analysis encompassing area under the blood plasma-As concentration time curve following zero correction and dose normalisation. In contaminated soil studies, As uptake into systemic circulation was compared to an arsenate oral dose and expressed as relative As bioavailability. Arsenic bioavailability ranged from 6.9+/-5.0% to 74.7+/-11.2% in 12 contaminated soils collected from former railway corridors, dip sites, mine sites and naturally elevated gossan soils. Arsenic bioavailability was generally low in the gossan soils and highest in the railway soils, ranging from 12.1+/-8.5% to 16.4+/-9.1% and 11.2+/-4.7% to 74.7+/-11.2%, respectively. Comparison of in vivo and in vitro (Simplified Bioaccessibility Extraction Test [SBET]) data from the 12 contaminated soils and bioavailability data collected from an As spiked soil study demonstrated that As bioavailability and As bioaccessibility were linearly correlated (in vivo As bioavailability (mgkg(-1))=14.19+0.93.SBET As bioaccessibility (mgkg(-1)); r(2)=0.92). The correlation between the two methods indicates that As bioavailability (in vivo) may be estimated using the less expensive, rapid in vitro chemical extraction method (SBET) to predict As exposure in human health risk assessment.