Characterizing the level and change of housing prices in cities is central to many empirical questions, whether prices are measured using rents or asset values. The task is complicated by the heterogeneity of the housing stock, the joint consumption of housing and neighborhood, and differences in accessibility. This paper focuses on intra-city location which, based on economic theory, is systematically related to housing prices. The final conclusion is that a sufficient statistic to describe both the level of and change in the average housing price requires that prices be aggregated from relatively homogeneous market areas and weighted by housing quantities such as dwelling units or interior space. Common repeat-sales and hedonic indexes are generally not weighted in this fashion but could be modified to do so.