The inclusion of spatial correlation of house price in hedonic pricing model may produce better marginal implicit price estimate(s) of the environmental variable(s) of interest. Most applications where a spatial econometric model is applied to the estimation of a hedonic property value model have used either a spatial lag model or a spatial autoregressive (SAR) error model. Incorrect spatial specification may produce even worse estimate outcome than OLS. Three issues regarding the specification of a spatial hedonic pricing model are considered. First, we question the "convention" of row-standardizing the spatial weights matrix. Second, we argue that the spatial error component (SEC) model is more theoretically intuitive and appealing for modeling house price. Third, we explore whether the choice of spatial model is important, empirically, using a large house sale dataset that includes measures of proximity to landfills. With one exception, estimated marginal implicit prices are fairly robust across all models.