The purpose of this study was to obtain empirical evidence regarding the effect on scores from three mathematics tests of (1) choice of test, (2) amount of training, and (3) recency of training. The findings were based on a random sample of candidates taking the April, 1948, series of College Entrance Examination Board tests. Groups were paired in a number of comparisons so that two of the factors under study were held constant while the third was varied. The paired groups were matched for "basic mathematical ability" by means of a regression technique which first tested the separate regressions for parallelism and then "corrected" the observed mean difference in test scores for any difference in ninth-grade mathematics marks. The results indicated that, on the average, students who had equal ability and who had just completed four years of mathematics courses made equivalent scores on the mathematical section of the Scholastic Aptitude Test and on the Comprehensive Mathematics Test. It was found, however, that candidates who took the Intermediate Mathematics Test made scores that averaged substantially higher than those of students of comparable training who took either of the other two tests.