This study identified factors that were potential predictors of student completion in a collegiate honors program, evaluated the relative predictive importance of these factors, and used them to describe how the completion groups differed. Records were examined from 336 freshman honors program participants at a large, Midwestern, public university. After five years, the students were classified into three completion groups: completers (n=62), partial completers (n=73), and noncompleters (n=201). Using an initial set of 16 pre- and post-college-entry variables that were grounded in Tinto's (1993) model of institutional departure, preliminary univariate and subsequent multivariate discriminant analyses resulted in completion-group prediction with 54.28% accuracy. The most important discriminating variables were the high school GPA, high school class rank, first-semester college GPA, gender, and initial housing assignment (honors housing or other). Implications focus on the improvement of completion rates through informed selection and support of honors program students. (Contains 3 tables.)