Willingness to pay for quality change may depend on heterogeneous perceived quality levels. In these instances, contingent valuation studies should include measures of quality perceptions as covariates in the willingness to pay model in order to avoid omitted variable bias. Variation in quality perceptions across respondents leads to a potential endogeneity of quality perceptions. We address the potential for endogeneity bias using an instrumental variable approach in which a measure of quality perceptions is included as a determinant of willingness to pay and is simultaneously determined by various exogenous factors. The willingness to pay model is estimated jointly with quality perceptions allowing for correlation of the error terms. Using data on willingness to pay for water quality improvements in the Neuse River in North Carolina we reject exogeneity of perceived quality. Correcting for endogeneity improves the measurement of willingness to pay by differentiating willingness to pay among respondents with heterogeneous quality perceptions.