This paper proposes a general empirical strategy to estimate willingness-to-pay (WTP) for exogenous risk mitigation when environmental risks are endogenous in protective actions and consumers are imperfectly informed about the ambient risk levels. The strategy consists of a set of survey techniques and the dummy endogenous variable model (Heckman, 1978) to control for correlation in unobserved errors that enter the WTP equation and the protection-decision equation. The method is applied to the non-market valuation survey data on arsenic contamination in drinking water. Our results indicate that the estimated WTPs are significantly higher for households without self-protective action. Our approach thus offers not only the correct welfare estimate for exogenous reduction of environmental risks, but also yields policy implications qualitatively much different from the conventional approach. We also estimate the welfare value of the policy to inform and educate the public about the arsenic risk simultaneously with public risk mitigation. The estimated welfare value is similar to, though slightly higher than, that of risk mitigation without information component. This occurs due to the competing effects of information dissemination and risk mitigation efforts. (PDF)