Increased risks of lung and bladder cancer have been observed in populations exposed to high levels of inorganic arsenic. However, studies at lower exposures (i.e., less than 100 μg/l in water) have shown inconsistent results. We therefore conducted an ecological analysis of the association between historical drinking water arsenic concentrations and lung and bladder cancer incidence in U.S. counties. We used drinking water arsenic concentrations measured by the U.S. Geological Survey and state agencies in the 1980s and 1990s as proxies for historical exposures in counties where public groundwater systems and private wells are important sources of drinking water. Relationships between arsenic levels and cancer incidence in 2006-2010 were explored by Poisson regression analyses, adjusted for groundwater dependence and important demographic covariates. The median and 95th percentile county mean arsenic concentrations were 1.5 and 15.4 μg/l, respectively. Water arsenic concentrations were significant and positively associated with female and male bladder cancer, and with female lung cancer. Our findings support an association between low water arsenic concentrations and lung and bladder cancer incidence in the United States. However, the limitations of the ecological study design suggest caution in interpreting these results.