Samples of household dust and garden soil were collected from twenty households in the vicinity of an ex-mining site in southwest England and from nine households in a control village. All samples were analysed by ICP-MS for pseudo-total arsenic (As) concentrations and the results show clearly elevated levels, with maximum As concentrations of 486 microg g(-1) in housedusts and 471 microg g(-1) in garden soils (and mean concentrations of 149 microg g(-1) and 262 microg g(-1), respectively). Arsenic concentrations in all samples from the mining area exceeded the UK Soil Guideline Value (SGV) of 20 microg g(-1). No significant correlation was observed between garden soil and housedust As concentrations. Bioaccessible As concentrations were determined in a small subset of samples using the Physiologically Based Extraction Test (PBET). For the stomach phase of the PBET, bioaccessibility percentages of 10-20% were generally recorded. Higher percentages (generally 30-45%) were recorded in the intestine phases with a maximum value (for one of the housedusts) of 59%. Data from the mining area were used, together with default values for soil ingestion rates and infant body weights from the Contaminated Land Exposure Assessment (CLEA) model, to derive estimates of As intake for infants and small children (0-6 years old). Dose estimates of up to 3.53 microg kg(-1) bw day(-1) for housedusts and 2.43 microg kg(-1) bw day(-1) for garden soils were calculated, compared to the index dose used for the derivation of the SGV of 0.3 microg kg(-1) bw day(-1) (based on health risk assessments). The index dose was exceeded by 75% (18 out of 24) of the estimated As doses that were calculated for children aged 0-6 years, a group which is particularly at risk from exposure via soil and dust ingestion. The results of the present study support the concerns expressed by previous authors about the significant As contamination in southwest England and the potential implications for human health.