This paper is concerned with objectively selecting and keying judgment items for maximum prediction of an academic criterion. A 40-item judgment test was constructed for entering college students. Populations for tryouts of the test were 106 first year law students at the University of Buffalo and 180 first year law students at Ohio State University. Keys were developed from a formula given for each law school. Only 30 items were used in the test. Table 1 shows the original and two experimental keys. Table 2 indicates the similarity between the various keys by presenting the percent of agreements among the total number of chances for agreement permitted by the particular items remaining in the key. Table 3 shows the validity of the original key and the Buffalo key as applied to the Ohio group. Table 4 shows the validity of the original key and the Ohio key as applied to the Buffalo group. (SGK).