The goal of this paper is the identification and estimation of the demand for environmental amenities and housing attributes as it is revealed by the housing choice of households located in the greater Zurich area. The precise and reliable estimation of the demand for housing attributes and environmental amenities is of prime importance in many environmental and housing issues. Whereas some questions like e.g. the impact of noise on housing prices can be addressed by simply estimating a hedonic model, the valuation of the welfare costs of noise pollution can only be addressed in an economic framework by the estimation of the demand for environmental attributes. In the paper we estimate a structural hedonic model in order to recover the willingness-to-pay for housing characteristics (lot size, living area, quality characteristics), location attributes (topography, neighbourhood quality) and major environmental amenities like the absence of traffic and aircraft noise. The estimation strategy follows the lines of a new methodology first proposed in an industrial organisation context by Berry, Levinsohn and Pakes (1995) that accounts for the heterogeneity of preferences. As in Bajari and Benkard (2002) we consider a setting with a continuum of choices. We are able to recover the distribution of preferences non-parametrically from a single cross-section by imposing restrictions on the utility function of consumers. The estimation is based on an set of 16'000 observations describing housing characteristics and the socio-economic traits of their residents located in the greater Zurich area. The database is provided by a mortgage originator and has been matched with a rich set of environmental attributes of the location through a GIS. The estimated preference parameters will be used in counterfactuals in order to explore important economic and policy questions. For example, we will be able to calculate a partial equilibrium decrease in house value that accompanies a rise in (aircraft) noise in some region holding the sociodemographic characteristics of this region constant. Furthermore, we will compute welfare changes for further housing attributes, like lot size, giving insights for land use policies aimed at protecting open spaces.