The results of the research program "Chemometrics" can be divided into sveral topics: i) Theoretical contributions. In this respect several new methods and concepts were introduced: a) new method for the evaluaton of the Kohonen maps' reliability of classifications, b) new method for mapping of multi-dimensional objects from an m-dimensional into 2d-space of 2-node error-backpropagation network signals. This new method is specialy interesting because it was achieved using a supervised learning method for solving a clearly unsupervised problem, c) new method for the representation of 3-d structures of chemical compounds. The new representation is, besides being uniform and unique, as well reversible. This means that from the new code icompounds 3d structure can be decoded directly. This property is specially important for the QSAR problems. Until our suggestion no 3d structure representatuon that would be uniform and reversible at the same time was known. (ii) Applicative achievements. Thee were too many to be listed here. In the four year period (1999-2003) we have published 26 papers in international journals (not counting full length articles published in various Conference Proceedings). Most of the published papers are based on the applications, be from the industry or resulting on the experimental work of various Master and PhD Theses. As a distinguished applicative success we can expose the publishing of the second edition of our book Neural Networks for Chemists and Drug Design (J. Zupan, J. Gasteiger). The success of the book is due mainly to the description and discussion of many applications in different fields of chemistry that is enabeled by the non-linear programing ability of the neural networks. In the second edition we have included several more examples elaborated in either our or in the Laboratory of Prof. Gasteiger from the University Erlangen-Nuerenberg. (iii) Through the entire period of this research program (1999-2003) our group has participated in at least one if not two Eurpean research projects (4-th or 5-th Framework of EU programs: Modelling of toxicity by neural networks, ImagetoX, VICIM). Additionally, we have carried out a number of bilateral collaborations: SLO-Bavaria, SLO, Italy, SLO-Croatia and SLO-Argentina. (iv) Educational achievements. In the reportetd period (1998-2003) in our group three (3) Diploma works, four (4) MSc, and three PhD Theses were guideded and accomplished.