Scores from 23 standardized tests that are used in application to graduate and professional schools are analyzed, primarily from the 1964-1982 period. The 23 examinations include tests of advanced achievement in 15 subject areas, along with tests of general learned abilities (the Graduate Record Examination/Verbal and Quantitative, the Law School Admissions Test, the Graduate Management Admissions Test, and Medical College Admission Test Reading and Quantitative Analysis subtests. Major conclusions include: (1) the quality of available data on test scores and on the background characteristics of test-takers is highly variable; (2) changes in test scores over a period should be measured in terms of standard deviation units, and not in points or percentages; (3) of 23 examinations, performance declined on 15, remained stable on 4, and advanced on 4--the greatest declines occurred in subjects requiring high verbal skills; (4) none of the basic demographic characteristics of the test-takers (age, race, gender, citizenship, or native language), in themselves, explain the observed changes in performance over the period; and (5) different undergraduate majors provide convincing explanations of observed changes in performance. Issues concerning the measurement of scaled test scores and the magnitude of change are addressed. Data on test performance are appended. (Author/SW)