When using a willingness-to-pay (WTP) format in contingent valuation (CV) to value water-shed restoration, respondents may protest by questioning why they should pay to clean up a pollution problem that someone else created. Using a sample selection interval data model based on Bhat (1994) and Brox, Kumar, and Stollery (2003), we found that the decision to protest and WTP values were correlated. Protest sample selection bias resulted in a 300 percent overestimate of mean WTP per respondent. Using different ad hoc treatments of protesters, protest bias resulted in moderate effects (-10 percent to +14 percent) after controlling for sample selection bias.