The report describes the research project that provided the basis for the GRE® Validity Study Service introduced in April 1979. The project involved 39 graduate schools and yielded more than 150 data sets, generally corresponding to departments, representing 19 fields or clusters of fields. The study pointed out problems of graduate-level validity studies caused by small samples, unstable parameters, restrictions of range, criterion selection, and so on, but demonstrated that useful results can be obtained despite such problems. It also demonstrated that the reliability of prediction can be improved by pooling data from corresponding departments across institutions.