A study was conducted to evaluate toenail arsenic concentrations as a biologic marker of drinking water arsenic exposure. Study subjects were controls in a US population-based case-control study of nonmelanoma skin cancer, randomly selected from drivers' license records (those < 65 years of age) and Medicare enrollment files (those > or = 65 years of age). Between 1994 and 1997, a total of 540 controls were interviewed and toenail samples of sufficient weight were collected from 506 (93.7%) of these. Beginning in 1995, a sample of tap water was taken from the participants' homes; a total of 217 (98.6%) water samples were obtained from the 220 subjects interviewed. Arsenic determinations were made from toenail samples using neutron activation analysis. Water samples were analyzed using hydride-generation magnet sector inductively coupled mass spectrometry. Among 208 subjects with both toenail and water measurements, the correlation (r) between water and nail arsenic was 0.65 (p < 0.001) among those with water arsenic concentrations of 1 microg/liter or higher and 0.08 (p = 0.31) among those with concentrations below 1 microg/liter (overall r = 0.46, p < 0.001). Our data suggest that toenail samples provide a useful biologic marker for quantifying low-level arsenic exposure.