Discrete choice models are used to elicit preference data from patients, medical and allied healthcare experts, and representative community samples. The resulting data are used in ecomomic evaluation studies to derive health utility values. In most reported studies, statistical methodology issues are usually glossed over and standard assumptions are made. However, statistical properties of discrete choice models present some interesting challenges to these more traditional views. OBJECTIVES: In this presentation we focus on two areas fundamental to the conduct of any discrete choice study: the use of orthogonal designs in experimental design and interpretation of model results, and highlight some misconceptions surrounding their current use. METHODS: More specifically, we highlight, with examples, that important properties of orthogonal designs assumed to underlie methods used in discrete choice studies don’t hold in general. RESULTS: We also discuss implications of applying the usual random effects or conditional models to discrete choice data. In both cases we discuss alternative approaches. CONCLUSIONS: The intention of this presentation is to inform researchers about these potential shortcomings in statistical methodology which is widely applied to discrete choice studies, and to encourage the development and use of alternative methods which may improve validity.