Affecting university rankings, school reputation, and financial well-being, student retention has become one of the most important measures of success for higher education institutions. From the institutional perspective, improving student retention starts with a thorough understanding of the causes behind the attrition. Such an understanding is the basis for accurately predicting at-risk students and appropriately intervening to retain them. In this study, using 8 years of institutional data along with three popular data mining techniques, we developed analytical models to predict freshmen student attrition. Of the three model types (artificial neural networks, decision trees, and logistic regression), artificial neural networks performed the best, with an 81% overall prediction accuracy on the holdout sample. The variable importance analysis of the models revealed that the educational and financial variables are the most important among the predictors used in this study. (Contains 4 figures and 3 tables.)