This paper considers the selection of appropriate functional forms for discrete choice models of housing market behaviour. It is argued that uncritical adoption of linear or log-linear forms may be inappropriate, and that Box-Tukey transformations are unnecessarily cumbersome and mechanistic devices in the search for more general functional forms. An alternative graphical assessment procedure based on partial residuals is first presented and is then applied to an empirical case study of tenure choice in London, England. Results are compared with a Box-Tukey analysis, and suggest that the graphical procedure constitutes an important exploratory modelling tool. Graphical functional form evaluation is thus seen to add important insights to analysis of the policy-dominated United Kingdom housing market.