In response to the perceived quality of a non-market good, households may choose to incur avertive expenditures as a substitute to its aggregate provision, thereby revealing an (inverse) demand function. When unobserved heterogeneity affects both perceived quality and avertive behavior, identification is plagued by a problem of endogeneity. In this paper, I propose to use a first stage model of perceived quality as a function of objective quality to address the endogeneity problem. The exclusion restriction is theoretically motivated, and requires people to have well-formed perceptions of the quality of the good. Moreover, the first stage quantifies the relationship between perceived and objective quality measures, so that this approach can provide micro-consistent marginal willingness to pay estimates for changes in the policy-relevant objective level of provision. As an illustration, I employ data on avertive expenditures for two qualitative aspects of household tap water supply: water hardness and aesthetic quality in terms of taste and odor.