We present in this paper the advantages of the latent class vector model approach in the analysis of preference ratings. We illustrate the practical application of this methodology on the basis of a preference study of smoked salmons in European countries. A set of 30 samples was selected spanning the characteristic variability of cold-smoked salmon available in six European countries (Belgium, Denmark, France, Germany, Italy and the United Kingdom). Consumers from the main countries of consumption (Belgium, France, Italy, Germany and the United Kingdom) were asked to rate the acceptability of each product. A quantitative descriptive analysis with a trained panel and physical and chemical measurements of smoked salmons were also conducted. Using a latent class vector model approach, it was possible to identify different classes of homogeneous preferences. Each of these classes was characterised by the country of origin of the consumers, sensory descriptions and the physical and chemical properties of the products. (C) 2007 Elsevier Ltd. All rights reserved.