A methodology is presented that incorporates site characterization information (in the form of hydraulic conductivity measurements) into a health risk assessment to individuals utilizing contaminated household water developed from groundwater sources. This methodology explicitly incorporates uncertainty (including the statistical parameter uncertainty arising from a limited number of hydraulic conductivity measurements) and variability (such as the variability in individual physiology) in all parameters, resulting in a nested Monte Carlo procedure. The methodology is demonstrated using a hypothetical case study where the sensitivity of various end points (including human health risk) to increased sampling of hydraulic conductivity are explored. It is shown that an increase in the number of measurements of hydraulic conductivity decreases the error in and uncertainty of predicted human health risk. Cost-benefit curves are developed and incorporated into a simple decision framework to determine the number of samples of hydraulic conductivity that are needed to accurately predict the human health risk from a contaminated site. ¿ 1999 American Geophysical Union