The use of student ratings of college instructors has steadily increased, with an attendant increase in the use of these student ratings in decision-making related to merit increases, promotion, tenure, and institutional severance. While a substantial body of research on student rating of instruction exists, the ambiguous or actually conflicting results of several of these studies has also led to concern by many professionals about the functional utility of student ratings. Using a sample of nearly 2,000 courses offered at the University, comparisons were made of: (1) correlations between Global Instructor Rating (GRI) and static course and student characteristics; and (2) predictor variables, order, regressions of student and course characteristics on GRI. The study collected one of the largest and most comprehensive sets of data on the subject of student evaluations of teaching. On the basis of the analysis it appears that, at a minimum, only a rather small portion of the total variance in instructor ratings can be attributed to demographic characteristics over which they have little control. (Author/KE)