This study investigated the relation between the entering characteristics of the 1,210 students who entered the seven residential teachers’ colleges in Jamaica in September 1976 and their subsequent performance in college. There were 41 predictor variables (encompassing biographical data, previous education and work experience, and general and specific performance in public examinations) and 33 criterion variables (specific and average grades in compulsory and optional examination. subjects, teaching practice, internship and individual study). The data were analysed using correlations and a series of multiple linear regression analyses in which poor predictors were successively eliminated. It was found that the examination average grade could be most efficiently predicted using a Student’s number of O-level passes and the scores on either the English and Mathematics or the Learning Potential entrance examination, giving a multiple correlation of about 0.50 in all courses. Performance in English and Mathematics was best predicted using the corresponding entrance examination score, but performance in other college examinations and in teaching could not be reliably predicted using the data collected. Implications are drawn concerning the selection of students for the restructured teachers’ college course recently introduced.