Data from a population-based repository in Manitoba showed that students who are male, young for grade, and in Grades 1, 2, 7, and 8 were the most likely to be retained. After controlling for key student factors including socioeconomic status, school changes, and key school characteristics including stability of the student body, retention was a strong predictor of high school withdrawal rates. The odds of student withdrawal were 3 times higher for students who have been retained once and almost 8 times higher for students who had been retained more than once. Not only do withdrawal rates suggest that retention is potentially harmful, but also Grade 3 standards test results taken in consecutive years by retained students indicate that only one in four retained students improve their score. These results generally replicate the research done in the United States. (Contains 5 tables and 3 notes.)