The paper reconsiders the theoretical motivation for studying the housing market by means of the concept "housing services." The analysis uses a model of consumer behavior that takes into full account the heterogeneity of housing. An assumption about the hedonic price function allows the researcher to define a scalar indicator of housing quality, that can be interpreted as the quantity of housing services available at a constant price per unit if a second assumption about the hedonic price function is made. The central result can be considered as an extension of Hicks' composite commodity theorem to cases in which expenditure on a bundle of characteristics is a nonlinear function of the quantities of these characteristics. The assumptions needed for aggregation impose testable restrictions on the hedonic price function, and a procedure for testing their validity is proposed. The basis of this approach is an econometric method for estimating functions with discrete explanatory variables without imposing assumptions about their functional form. This allows one to test the validity of the aggregation conditions essentially without joint assumptions. The method is applied on the basis of data for the two largest Dutch cities. The conditions for a constant price per unit of housing services are violated. Aggregation of bundles of housing characteristics into a scalar measure, which may alternatively be inter_preted as a quantity of housing services available at a varying price per unit, appears to be possible for the data used. However, the concept housing services loses much of its convenience if it cannot be used in combination with a constant price per unit.