Market segmentation often requires that different sets of variables are taken into account. Examples are demographic, psychographic or behavioral variables. A combination of complicated techniques (e.g. canonical analysis, cluster analysisis and discriminant analysis) is needed to obtain results that can be evaluated and implemented. To combine such techniques has some technical drawbacks; the main drawback, however, is that the procedure is difficult to communicate. Therefore, the need arises of simpler techniques that can do the same job. If possible, it is preferable that the complete segmentation process is based on a single technique. Latent Budget Analysis is a decomposation techniques for two-way tables that also can be used for segmentation purposes. It requires that the data can be written In the form of budgets (in terms of behaviour, time or money). The budgets can be related to single individuals, demographic or psychographic categories. The results of Latent Budget Analysis are 'factors' which have a simpler interpretation than the results form ordinary factor or canonical analysis. Moreover, they have a simple geometrical representation. This representation can directly be used for segmentation purposes. In the paper some general considerations in market segmentation are discussed. The basic theory of Latent Budget Analysis is given (leaving out most technicalities). Latent Budget Analysis is applied to the market of non-alcoholic and light alcoholic beverages. It Is shown how individuals can be clustered Into segments, and also how units of products can be attributed to different market segments.