Rather than studying the structural paths through which variables affect student persistence in education, this paper offers a reduced form model that focuses on precollege, demographic, and certain current achievement and financial aid variables. This approach does not specify structural paths, but it does have the advantage of requiring only information available in student records. The empirical model used is the discrete time hazard model because duration data are collected term-by-term until student stopout. Stopout, for this study, is defined as the first occurrence of non-continuous enrollment. The model is illustrated with data for all 3,556 students entering the University of Minnesota Twin Cities campus as New High School students in 1985. These students were observed post hoc for 22 terms (3 terms per year for just over 7 academic years) to study student exits from the institution. The basic model assumes time-constant effects of the independent variables believed to affect stopout. Results are generally consistent with those from other studies of persistence, but show more time profile detail. Race differences were found to vary over time, but these differences might reflect differences in the institution, the students, or inadequate multivariate controls in other studies. The hazard model permits analysts to examine whether factors influencing student dropout or graduation vary by initial year of enrollment. More detailed analysis of subgroup hazards should help administrators meet the needs of at-risk groups. (Contains two figures and five tables.) (SLD)