A study assessed the effectiveness of various ways of identifying at-risk middle school students for participation in a dropout prevention program. It was considered important to explore factors related to identifying students who were most in need of the program. During the first semester of the project, teacher recommendations, attendance, age, and a variety of perceived competence measures were used as predictors of failing grades, grade point average, and frequency and severity of discipline referrals. The study took place in a small-city school district in North Carolina which had a dropout rate above the state average. Contrary to national trends, the typical dropout in the district was a white male. Subjects were all 334 sixth-grade students. Findings confirmed that it was possible to predict with a fair amount of accuracy which children were at risk of receiving failing grades and what their grade point average might be. The number of failing grades appeared to be a theoretically strong outcome indicator. It is concluded that teachers can identify early in the semester students who are at risk of failing, and they can do so with measures that are easily accessible and fairly accurate. (RH)