Chronic arsenic consumption can cause vascular diseases. Adverse vascular effects of arsenic in drinking water in the United States have not been studied. This study investigated the ecological relationship between the population-weighted mean arsenic concentration in public drinking water supplies and mortality from circulatory diseases in 30 U.S. counties from 1968 to 1984. Mean arsenic levels ranged from 5.4 to 91.5 micrograms/l. Standardized mortality ratios (SMRs) for diseases of arteries, arterioles, and capillaries (DAAC) (ICD 8th/9th revision, 440-448) for counties exceeding 20 micrograms/l were 1.9 (90% confidence interval [CI] = 1.7-2.1) for females and 1.6 (90% CI = 1.5-1.8) for males. The SMRs for the three subgroups of DAAC--arteriosclerosis, aortic aneurysm, and all other DAAC--tended to be elevated. With respect to the same arsenic group, the SMRs for congenital anomalies of the heart (ICD-8/9, 746/745-746) and circulatory system (ICD-8/9, 747) also tended to be elevated. Two competing interpretations emerge as possibilities: either there are spurious associations resulting from invalid outcome data or causal associations.