When applying the contingent valuation method (CVM) in risk reduction studies, some studies report that willingness to pay (WTP) is insensitive to the magnitude of risk reduction while other studies do not. On the other hand, social-psychological research has shown that the affect heuristic biases judgments on probability, relative frequency, and risk. This article examines both magnitude (or scope) effect and affect heuristic (or representation) effect on WTP for risk reduction measures against tsunamis by introducing two experimental (i.e., absolute and relative) systems with four different representations to evaluate public behaviors in two different scenarios of risk reduction. Two common denominators (100 and 10,000) are introduced into absolute risk reduction representation (i.e., "of every 100 persons, from present 2 deaths to 1") to form different formats (i.e., "of every 10,000 persons from 200 to 100," and "of every 100,000 persons from 2,000 deaths to 1,000"). There is little evidence that WTP estimates are sufficiently sensitive to the magnitude of risk reduction, but relative risk reduction representations may be better than the absolute one given in CVM mail surveys when the risk is small. There is a statistically insignificant effect of risk reduction representations on respondent frequency, but mixed effects on the monetary values of WTP at the level of 0.05. The representation effect of absolute risk reduction on the WTP value varies with the common denominator. The larger the common denominator, the less the WTP to reduce the risk of tsunamis, and the significance probability is improved to less than the level of 0.05 when the common denominator becomes large enough. The findings suggest that improved methods are required for estimating the rates of tradeoff between fatality risk and other goods among consumers.