A salary prediction model for college faculty that is used at Georgia State University was reviewed and tested using multiple regression analysis. Various model specifications, incorporating academic rank, academic discipline, and academic experience, including professional and personal background characteristics, are reviewed. Academic rank is an important determinant of salary, and it is strongly related to three key indicators of academic performance: teaching ability, scholarship, and service. Academic discipline is another important salary determinant that is influenced by the supply and demand in the academic labor market for that discipline, institutional factors, including fluctuation in student enrollment; and department factors which influence salary, including paradigm development. Paradigm development is the level of agreement within an academic discipline of research methodology, course content, and research issues and problems. Background characteristics of faculty interact in a complex fashion with academic discipline and department to influence salary outcomes. The next most important characteristic in model specification is a component reflecting academic experience. The prediction equation incorporates rank, administrative status, discipline, and years in rank, endowed chair and contract status, or full-time equivalence. To explain the dynamics of faculty rewards, other equations were also employed, which include possible discriminatory variables, such as sex, race, age, and years employed. The equations were developed for the major colleges of the university, and separate equations for males and females were calculated. It was found that 90 percent of the variance in the salaries of teaching faculty could be accounted for by the prediction model. (SW)