In this paper we consider a first application of the Learning Vectorial Quantification Neural method (LVQ) to the problem of studying and distinguishing between different populations within an stellar catalogue of the solar neighbourhood (a complete description can be found in Hernández-Pajares and Monte, 1991, Artificial Neural Networks, Ed. A.Prieto, Lecture Notes in Computer Science 540, Springer-Verlag, p.422). It consists, briefly, in the approximation of a set of vectors in a certain characteristic space that contains continuous elements. The representative points for every cluster are the centroids, calculated in such a way to minimize the distortion. Each of those can be labeled with integer numbers using a 2D representation that preserves the neighbouring property in the characteristic space: the Kohonen Map (Kohonen, 1988, IEEE Computer, 21 Nbr.3).