The limited success in predicting academic achievement by the use of personality factors may be attributable to failure to delimit sufficiently the nature of the group being predicted, resulting in subtle differences lying hidden in the remaining random variance of the procedures. This study attempted to enhance prediction by a more precise identification of the subjects for whom achievement is being predicted. A sample of 250 college freshman males was used to develop a prediction schedule. Predictions of academic success, based on the criterion of grade point average (GPA), were made on the basis of a clinical profile interpretation of the four clusters of scales on the California Psychological Inventory (CPI). Predictions were made by a gating process through which the predictions arrived at in each succeeding cluster of scales were used to refine the prediction level assigned in the preceding cluster(s). An attempt was made to quantify this process of clinical prediction using an experimental group of 20 and a replication group of 10. The results of the experimental group were analyzed and a correction formula developed. The distribution of predictions and G.P.A. were skewed in both groups. Prediction in the experimental group was improved by the correction factor but not in the replication group; the results are enigmatic and further investigation is necessary. Some possible explanations are suggested. (LR)