The widespread use of student evaluations to rate faculty has raised the question of whether high student evaluations can be gained simply through the process of faculty giving higher grades to students, or whether learning of students is a critical factor in such evaluations. Four different models were tested which represented different relationships between students= expected student grades and student evaluations of the quality of instructors, with and without student motivation, ability, and amount learned as potentially important variables. Evaluations from 119 students of four different instructors were used for the data set. Statistical tests of the alternative models indicated that a more complex model incorporating student motivation and ability levels as factors affecting student evaluations of instructors provided the best fit to the data. The fit was superior to that of a model using only expected grades and student evaluations of instructors, indicating that students= evaluations of faculty did not appear to be based solely on the grades students expected to receive. The complex model also fit the data better than a simpler model using only perceived amount learned, expected grades, and instructor ratings. For this data set, instructor ratings were not simply a function of expected grades, or simply a function of perceived amount learned, but a function of motivation, ability, amount learned, and grades.