Little has been done to analyze the forces that predict whether high-risk students will stay in school or drop out. Current and retrospective data from 27 high-risk youths were studied to shed light on this problem and to suggest ways of improving educational services. The at-risk students were from low socioeconomic backgrounds, were Mexican American, and were language minorities. Results showed that the high-risk youths did not differ significantly on standard school entry-level variables. However, more and regular exposure to learning activities in the home and school, in Spanish as well as English, seemed to be associated with greater educational attainment and achievement. The only noticeable differences in elementary school between eventual dropouts and college-bound students appeared on tests of reading achievement. College-bound students and graduates reported more positive memories of elementary school. Dropping out was not influenced by employment, peer relationships, or school attendance policies. But by the end of eighth grade, eventual dropouts could be distinguished by their lower grades and poorer attendance records. There was no evidence that tracking, i.e., locking students into academic instructional streams, influenced dropping out. The findings suggest three general areas for interventions to improve services: (1) academic tutoring and programming; (2) entry considerations, for both elementary and junior and senior high school; and (3) counseling, school participation, and school structure in general. (JAC)