The research methodology comprises theoretical, empirical and evaluation stages.  The theoretical stage provides evidence that substantiates the need for the study and outlines possible ways to address spatial elements in hedonic price modelling.  The empirical stage illustrates the application of GIS and spatial statistics in the estimation of hedonic models for housing markets in Glasgow, Scotland, using 2,715 house prices for 2002 and 61 independent variables.  GIS is used in this study to construct spatial variables including detailed accessibility measures, to help detect the hedonic problems of heteroscedasticity and spatial autocorrelation, and for visualisation.  Spatial statistics are used to test formally and model explicitly the spatial autocorrelation.  The evaluation stage assesses 46 hedonic models, using OLS and spatial hedonic, for a priori segmentations involving the spatial, structural and nested sub-markets.  It also draws general conclusions about the importance of detailed accessibility measures and spatial statistics in sub-market modelling. This study finds that the nested sub-market modelling using a spatial hedonic approach is most effective, followed by the spatial and structural sub-markets.  The OLS sub-market modelling generally reduces spatial autocorrelation but does not eliminate it.  There is a greater incidence of spatial autocorrelation when the market size, with measured by geographical area or density of dwellings is larger.  The spatial hedonic modelling improves the performance of the individual OLS models and the three segmentation approaches, although the relative performance of the latter remains unchanged.  Nevertheless, will the spatial hedonic, the entire market model outperforms the OLS model of structural sub-markets.  Flat-based OLS sub-market models benefit substantially from the spatial hedonic.  The results also suggest that an individual accessibility measure is more significant than the zonal measure because it is able to capture the micro effect of location on price.  Further, spatial statistics produce more accurate, robust and reliable estimates of implicit prices.