A system to identify the potential dropout was studied at a two-year urban institution. All entering freshmen participating in a compulsory freshmen orientation course were administered a biographical inventory during the fifth week of enrollment. An academic or scholastic index was derived, based on high school averages and basic skills placement scores in reading, writing, and mathematics. The spring 1981 group included 328 students, while the fall 1981 group involved 704 students. At the beginning of the second semester, all students were classified into one of four groups: dropout, failing index; dropout, passing index; returnee, failing index; or returnee, passing index. In addition, the spring 1981 cohort was reclassified at the beginning of the third semester. Each group was divided into those who had earned satisfactory academic indices and those who had not. Discriminant analyses were conducted using two, three, and four group criteria. After analyses were carried out on the total cohorts, the data were divided into four major curriculum areas: liberal arts, science and technology, nursing and medical laboratory technology, and business. A separate group of students was evaluated to cross-validate the proportion of correct group placements of known placements against a group of test students for whom placements were unknown. It is concluded that the system should not be implemented as it currently exists, and problems that prevented greater predictive success are briefly considered. Tentative findings suggest that the most important factors involved in student attrition may be related to ethnicity and language primacy, student commitment to educational goals, college workload, scholastic readiness, and ethnic group and marital status interactions. (SW)