Regression analysis was used to study retention and graduation for the fall 1996 entering class of students at a midwestern research extensive university (n = 3,610; 44% female, 8% minority, 77% in-state). Logistic regression was used to predict the likelihood of a student being retained for each of four years, and the outcome of graduation at the end of years four, five, and six. Odds ratios were employed to provide insight into the relative contribution of demographic characteristics (age, gender, ethnicity, in-state residency), ability (high school rank, high school rank[squared], ACT score), environmental (university athlete, university honors program, first-generation student, entering college), and financial aid data (gift, loan, and work-study) characteristics. (Contains 2 tables and 1 figure.)