This study investigates the temporal stability and transferability of dichotomous choice willingness to pay responses and their determinants from two large-scale contingent valuation surveys in the area of flood control and wetland conservation. The study considers a time period between surveys which is more than double that considered in previous test-retest analyses. Whereas such previous studies have reported stable values over relatively short time periods, the present study finds a statistically significant decrease in real willingness to pay over this more extended time period. Analyses of model transfer between the two survey periods indicate that models derived solely from economic-theoretic determinants pass transferability tests. However, expanding these models to include more ad hoc, transitory factors yields nontransferable models. This provides a guide for future analyses.