Self protection and altruism are crucial behavioral factors in determining the effectiveness of public policies aimed to improve human health from environmental hazards. This paper examined people’s arsenic mortality risk perception in the drinking water for themselves and their children using the Bayesian learning framework. A two-stage structural model within the random utility framework was developed to model the household’s risk averting behavior with respect to arsenic-related mortality risk. The empirical results indicate that parents engage in a form of mixed altruism. Parents are willing to spend more to make a trade-off between their risk and their children’s risk.