This paper describes a recently developed method for predicting market shares of products in a competitive market from qualitative judgements of a small sample of respondents. This "conjoint choice approach" is an extension of traditional conjoint analysis. The conjoint choice approach uses a different data collection procedure, in which stated preferences of available products and consumer choice between competitive products are central. Data analysis involves the calibration of logic models and the derivation of weighting factors to make proper aggregate predictions. The result of the analysis consists of a simulation tool, which takes the form of a computer program for IBM-compatible PC's. The paper presents the approach, and illustrates it by giving an example of its application to rail travel on the Plymouth-London corridor. It concludes by indicating some planned further extensions and applications.