Willingness to pay (WTP) estimation typically involves some strategy for mapping nondichotomous contingent valuation (CV) responses onto a dichotomous yes/no dependent variable. We propose a new approach to selecting which responses qualify as ‘yes.’ We apply the proposed method to polychotomous CV data for preventative land management programs in the Great Basin. We also estimate WTP using other methods of response recoding found in the literature. By contrasting the results under different approaches, we demonstrate how and why WTP point estimates vary across recoding methods and discuss the comparative advantages of our more generalized recoding approach that is based on predicted probabilities of ‘yes’ responses.