Hedonic price models have been broadly employed to explore the factors of housing prices changes. Especially market-wide hedonic models have tended to provide preliminary investigation of spatial distribution of housing prices. This paper seeks to explain the effectiveness of different segmented house price structure in Istanbul. Three different approaches are used in order to model the housing price differences. First, a standard market-wide model is employed which is taken as a reference point. Second, a hedonic price model that includes neighbourhood dummy variables as a proxy for segments within the model is used. Third, a multi-level model which includes segments and their interactions with each other is employed. Comparing the performances of these three approaches, this paper concludes that taking housing market segmentation into account improves the effectiveness of the models.