ABSTRACTA set of terms in descriptive analysis should enable differentiation among products, specification of the sensory properties of the products, and sufficient characterization to permit its recognition or identification. Meeting these criteria enables the use of descriptive data in several applications: (1) interpretation of other sensory data, (2) correlation with instrumental measures, (3) quality monitoring, and (4) product development and maintenance. Several aspects of the specific terminology and the way it is taught are important in effective descriptive analysis. Terms should be uncorrelated with one another. Terms should be related to physical or chemical references. Training a panel requires the establishment of a system in which relevant dimensions can be abstracted and attended to. In this paper, specific problems of descriptive analysis training such as the choice of reference sets and the boundaries of terms are discussed.