For both arsenic and nitrate we developed several summary measures. As expected, there was variation in how well each correctly ordered households with regard to actual levels. More important, however, were the radius (maximum distance) between the home and sampled water source, and whether the home was on a water system. Precision increased as the radius decreased, but ability to link homes to any sampled water source also decreased. This effect was sufficiently pronounced that if agreement became marginally acceptable (ρ > 0.40), "participation" percentages ranged from poor to marginal (30% for arsenic, and 50-72% for nitrate). Precision was also greater for homes off, rather than on a water system. Perhaps for homes with private wells, the spatial relationship with its actual water source is more geographically based than for homes served by a water system. This may be particularly true for those relying on surface sources, which may be quite distant from their ultimate tap destinations. Contaminant levels measured at the source may more likely reflect levels at the tap in homes not on systems, i.e. without an intermediate supplier that may treat and mix water to meet water quality standards. Although the number of homes in our study on private wells was very small, and our GIS-based estimates may have benefited from state-mandated monitoring that more than doubled the number of monitored sources included, it is encouraging that precision appeared greatest for this subset of homes for which it would be impossible to apply the more traditional approach of linking water quality records according to which system served the home.