The purpose of this longitudinal study was to examine associations among background characteristics, commitment levels, institutional involvements, and persistence for male and female university students. Data were collected from college records and surveys completed by 316 university freshmen at the beginning of and 10 weeks into the 1984 fall semester and at the beginning of the 1985 spring semester. Logistic regression analysis in conjunction with LISREL was used to analyze the data. Within LISREL, an initial model was specified using paths described in Tinto's model of college student attrition. Tinto's model consists of six key constructs linked causally: background characteristics, initial commitments, academic and social integration, later commitment, and persistence. Results reinforced Tinto's conjecture that background effects do influence persistence directly. Initial levels of institutional and goal commitments strongly influenced later levels of the same variables. For both males and females, academic and social integration positively influenced persistence. For females, academic integration, social integration and later institutional commitment were significant predictors of persistence. For males, only academic integration and social integration significantly and directly influenced persistence. (JAZ)