The utility of multiple discriminant analysis in predicting career interests in six specialities of engineering--aeronautical, chemical, civil, electrical, industrial and mechanical engineering--was studied. Freshman engineers at Purdue University were administered the Purdue Interest Questionnaire (PIQ). Multiple discriminant analysis was performed on the six Engineering Specialty Scales of the PIQ using the obtained scores of 292 students. These students were then classified by means of discriminant functions into six engineering fields of study. Each predicted classification was then compared to the actual, fourth semester classification, and accuracy percentages for each field were computed. As a cross-validation of the discriminant analysis, these same discriminant functions were used to classify in the same manner a second sample of 301 students. Then the two student samples were pooled together, a new discriminant analysis was performed, and by means of the new functions the group of pooled students was again classified as before. In general, correct classification percentages significantly exceeded the percentages predicted by chance. Finally, it was observed in the cross-validation procedure that consideration by the counselor and a student of the first-or-second classification choice as optimal for success greatly increased the likelihood of making the best decision. (Author/CTM)