The implementation of a molecular similarity algorithm in a CNDO-like approach has been performed owing to the fact that the information provided by such a procedure does not need great refinement. The present version of MOLSIMIL-88 has been written in FORTRAN-77 for use in IBM-PC or Compatible XT/AT machines with math co-processor. The code has been designed to be compatible and transferable to any machine with an adequate compiler facility. Program dimensions can be changed with minimal editing work. The ASCII source code as such may be compiled and runs without errors on any VAX computer with a VMS operating system. The program needs a minimal amount of input data for a given molecular family, consisting of: coordinates, atomic numbers and gross atomic populations. Appropriate simple input parameters stored in a separate file provide the program with the necessary hints to perform various optional runs using many choices of gradual complexity, while conserving the molecular family structure input. The essential computation output is formed by a density correlation matrix between any given couple of molecular structures of some molecular set. The program uses the correlation matrix elements as a starting point to compute trees and graphs in order to obtain complete visual information on the whole molecular set.