This study applies stated-preference methods to estimate the public good values of preservation and restoration of a local historic landmark in a medium-sized US city. The survey device centres on a referendum-style dichotomous-choice question regarding city participation in a restoration partnership. Use of a double-split sample allows analysis of the effect on valuation of both heritage information and willingness-to-pay versus willingness-to-accept constructs, where the latter is approached by the method of paired comparison, allowing a non-substitutability effect to be separated from any endowment effect. Econometric analysis using a standard binary logit model indicates the existence of a strong non-substitutability effect and a significant information effect that further suggests the importance of non-substitutability in valuation.