Discrete choice modelling is one of the main tools used in transport research and well-designed experiments are central to obtaining a richer understanding of individuals’ decision processes. Traditionally, designs considered principles of orthogonality, balance and no-dominance, but those designs were not necessarily efficient. In order to extract the maximum information from the data with fewer resources, recent designs minimise certain properties of the asymptotic variance-covariance (AVC) matrix. The design principle is to obtain the most reliable parameter estimates for a predefined error structure (Rose and Bliemer 2005a,b, 2009; Bliemer et al. 2009a). The efficient design of stated preference surveys relies on prior parameter estimates typically sourced from pilot studies. This paper investigates the use of hedonic regressions on readily available market data as a source of prior estimates. Secondly, the efficient design is a result of a search within a large space of potential designs. Thus, the second aim of this paper is to present the use of genetic algorithms to aid in this search.