The theoretical and empirical implications of information effects on estimates of existence values (EV) are analyzed using a Household Production model. If households operate in perfect information markets, current practices of providing "policy relevant" or "complete" information to _respondents to Contingent Valuation (CV) surveys may produce larger benefit estimates than should be used in damage assessment and benefit-cost analysis. However, the probable failure of information markets provides a rationale for the provision of information. An empirically applicable rule is derived that determines the optimal level of information use when estimating existence values. Adopting this methodology would provide an improved decision rule, reinforce the validity and credibility of CVM results, and promote the acceptance of EV and other nonuse values as a desirable component of benefit-cost analysis.