The risk of arsenic exposure to deer mice (Peromyscus maniculatus) living in areas of naturally and anthropogenically elevated arsenic levels was determined using three separate calculations of arsenic daily intake: Estimated daily intake (EDI), bioaccessible EDI (BEDI), and actual daily intake (ADI). The present work is of particular interest, because the risk assessments were determined for animals naturally exposed to arsenic. Gastric fluid extraction was used to obtain bioaccessibility data for soil and plant samples collected from three study sites (background, mine forest, and tailings) in Yellowknife (NT, Canada). Calculations using the EDI indicated that deer mice living in tailings habitat (average soil arsenic concentration, 1,740 +/- 2,240 microg/g) should have been experiencing serious health effects as a result of their exposure to arsenic. Using BEDI and ADI in the risk assessment calculation, however, resulted in an order-of-magnitude decrease in calculated risk. In addition, results calculated using the BEDI and ADI were not significantly different, suggesting that using bioaccessibility provides a more realistic estimate of potential risk. The present results provide evidence that the use of EDI in traditional risk assessments may seriously overestimate the actual risk, which in some instances may result in expensive and unnecessary clean-up measures.