One of the best ways of structure elucidation would be if a system could select the most probable structure from a gigantic file including all the possible structures that are known to exist, or that might exist from a chemical point of view, with the help of structural information, for example, a chemical spectrum. However, it is not possible to store all possible structures in a file. For example, even CUH4 has 5 731 580 isomeric structures. To overcome this, the present CHEMICS is designed to store all the substructures (called “components”) necessary for building any likely structures. The set of components has been devised so that it is possible to construct any structure by selecting appropriate components from the component set. To store such a set of components in a computer is logically synonymous with storing all the complete structures that could be present. CHEMICS (CHEMICS-6) contains 189 components for the structure elucidation of organic compounds consisting of only C, H, and 0 atoms. A trial and error method was adopted in the selection of these components, with due regard to the prerequisites that the components should have no substructures overlapping one another and that the presence of the components could be deduced from structural information such as molecular formula and spectral data. Furthermore, 572 components have recently been prepared for CHEMICS-7 to handle samples that contain N, halogen, and S atoms in addition to C, H, and 0. In the image space, all components are possible for an unknown compound (the analyst faces an almost infinite set of possible structures when he is without any structural information). CHEMICS-6 and -7 have the following tasks: (1) to eliminate unappropriate components, from the prepared component set, that are inconsistent with the molecular formula and spectral data; (2) to generate complete structural formulas from the retained components; (3) to exclude unlikely structures from those that have been generated; (4) to generate possible stereoisomeric structures, if a candidate structure possesses a stereocenter; (5) to output the most likely candidate structures, ideally the single correct solution. This paper describes the progress story up to the present CHEMICS. As to the details of the current CHEMICS, readers are requested to refer to the article that appeared in Computer Enhanced Spectroscopy (1983, I , 5 5 ) .