A model of student attrition was synthesized from psychological, sociological, and educational sources, and contains six sets of variables: background, organizational, personal, environmental, attitudinal, and intent to leave. The model was tested with 1,909 full-time and unmarried university freshmen at a major midwestern university. The sample was divided into four groups based on the student's sex and level of self-confidence, and multiple regression and path analyses were used to analyze the data from the study instrument. Background variables included father's education, mother's education, performance in high school, high school and home town size, and distance to home. Organizational variables included university grades, informal contact with faculty, centralization, memberships in campus organizations, finding the academic program competitive, courses, and absenteeism. Personal variables were goal commitment, major and occupational certainty, and confidence. Environmental variables included opportunity to transfer, likelihood of marrying, ease of financing one's education, and family approval of the institution. Attitudinal variables were loyalty, certainty of choice, satisfaction, and practical value. It was found that intent to leave and university grades were the best predictors of attrition; high confidence compensates for absenteeism and low grades in reducing dropping out. The correlation coefficients ranged from .43 to .53. For each of the four path analyses (high/low confidence women and high/low confidence men), intent to leave showed a consistently high positive relationship with dropping out, while university grades were negatively related to dropping out. Recommendations are presented, and the way that this model differs from those of Spady (1970) and Tinto (1975) is considered. (SW)