"This study investigates the value of improved urban land in a North American context, more specifically in Montreal (Canada). It follows two main goals in response to the difficulties of: * Explaining the improved urban land value determinants; * Estimating its market value. // These interrelated difficulties, particularly regarding ìReal Estate Evaluationî and ìUrban Economicsî fields, are specifically engendered in a shortage of vacant land market. In fact, in the current context of cities, the majority of the observed prices in the marketplace include the price of the land and that of the buildings. In such a situation the prevalent question is to know how much the price of each represents in the total selling price of the properties. In the first goal of the research, different theoretical conceptions explaining the formation of land prices are examined. Their analysis makes clear that the land price is usually confounded within the total property price. In the second gaol research, this one empirical in nature, available methods of property evaluation are explored. Their synthesis demonstrates that they remain inoperative in the case of improved non-income generating urban lands, allotted especially to the single family residential properties. Income-producing lands, however, find satisfactory answers within the residual rent method, supported by the classical ricardian rent theory. Contrary to the vacant land market, the single family properties market contains a sufficient number of comparable sales, but encounters the difficulty of total price apportionment between the land-building components. This study perceives that it can nevertheless be treatable under the neoclassical view of utility, by using the hedonic method it supports. It is accepted that a hedonic method allows for the ""decomposition"" of the total property price between its multiple attributes and the reconstitution of its total value by the sum of their marginal contributions. Although frequently applied in this matter by academicians and practitioners, its capacities have not been explored with the objectives of explaining and estimating ìseparateî values for land and buildings. In its present form, it does not discriminate, neither in theory nor in practise, the particular attributes of these components. This research therefore suggests the incorporation of a new specification to the current hedonic models allowing them to deal with the two underlined difficulties. This new specification propose that the price of a property is not a sole ìbundleî of residential services as suppose hedonic models, but rather a function of two independent ìbundleî of services: * a first bundle containing the differential site advantages (DSA) ; and * a second bundle containing the differential housing advantages (DHA). // The proposed model explains the separate market value of improved lands through their DSA attributes, as is the market value of the buildings through their DHA attributes. It reaches also to an estimate of the ìseparate hedonic valuesî of land and building components; the sum of the two entities forming the total market value of a single family property."