The effectiveness of aquifer remediation is typically expressed in terms of a reduction in contaminant concentrations relative to a regulated maximum contaminant level (MCL), and is usually confirmed by sparse monitoring data and/or simple model calculations. Here, the effectiveness of remediation is examined from a risk-based perspective that goes beyond the traditional MCL concept. A methodology is employed to evaluate the health risk to individuals exposed to contaminated household water that is produced from groundwater. This approach explicitly accounts for differences in risk arising from variability in individual physiology and water use, the uncertainty in estimating chemical carcinogenesis for different individuals, and the uncertainties and variability in contaminant concentrations within groundwater. A hypothetical contamination scenario is developed as a case study in a saturated, alluvial aquifer underlying a real Superfund site. A baseline (unremediated) human exposure and health risk scenario, as induced by contaminated groundwater pumped from this site, is predicted and compared with a similar estimate based upon pump-and-treat exposure intervention. The predicted reduction in risk in the remediation scenario is not an equitable one - that is, it is not uniform to all individuals within a population and varies according to the level of uncertainty in prediction. The importance of understanding the detailed hydrogeologic connections that are established in the heterogeneous geologic regime between the contaminated source, municipal receptors, and remediation wells, and its relationship to this uncertainty is demonstrated. Using two alternative pumping rates, we develop cost-benefit curves based upon reduced exposure and risk to different individuals within the population, under the presence of uncertainty. Copyright ASCE 2004.