There is the potential for arsenic to enter the human food chain via ingestion by grazing animals. Data on the transfer of arsenic to ruminants have been too sparse to allow the development of dynamic models to predict changes in the arsenic contents of different tissues following ingestion. A study is described during which a group of 6-month-old lambs were given a single oral administration of 73AsCl3. Subsequently, concentrations of 73As in the tissues of groups of lambs slaughtered at intervals over a period of 181 days were determined. A true absorption coefficient of 0·46±0·055 (mean±S.E.) was determined which is considerably lower than expected from previous studies of non-ruminant animals which demonstrate complete absorption for inorganic arsenic. The resultant data were used to develop a compartment model to describe arsenic behaviour in sheep tissues. The derived model accounted for 80 % (n = 100) of the observed variation in the data. The model predicts that arsenic concentrations in tissues rapidly (< 40 days) reach equilibrium with the dietary intake level. Equilibrium transfer coefficient values (the ratio of the arsenic concentration in a tissue to the daily dietary intake of arsenic) for the important food-chain tissues were calculated as: (2·5±0·67)×10−3 days/kg for muscle, (9·1±1·96)×10−3 days/kg for liver and (1·1±0·14)×10−2 days/kg for kidney.