Abstract : This report covers the period from 1 June 1977 to 30 October 1980. A major task on this contract was to make a comprehensive review of the literature on software metrics and of quantitative measures of program testing. Another accomplishment was the development of an automatic procedure for testing FORTRAN programs with random numbers and random symbols. This procedure first drives the program with sets of random data, then senses the tracks taken, compares each generated track against all predecessors, then on the basis of the pattern of occurrence of new tracks, provides an estimate of the total number of residual paths. Additional sets of random numbers can be then generated. The program developed to instrument a given FORTRAN program and provide data for evaluating coverage is described in the Final Report of earlier work (MDC G6553, dated December 1976). Changes which have been made are described herein. In the related topics of Software Reliability, two methods of estimating the residual error content of an entire program on the basis of data obtained in the testing of portions of it have been developed and are detailed here. (Author)