A set of 65 evaluation questions was compiled from published instruments. A total of 200 undergraduate students rated both their best and worst teacher from the previous term. Each of the set of 65 items was rated twice for the best teacher and twice for the worst teacher, with order randomized. Factor analysis of ratings of the Best Teachers, the Worst Teachers, and the combined data resulted in similar factor structures. Seven evaluation factors were identified: Breadth of Coverage, Organization/Clarity, Interest/Dynamic, Difficulty, Learning/Value, Individual Contact and Interaction/Discussion. Reliability of individual items was determined by correlating student responses to the same item for the same instructor. Coefficients for the combined data varied between .61 and .97 with a median of .88. There were highly significant differences between mean ratings of Best Teachers and Worst Teachers on all items except those related to Difficulty. The clearest differentiation was for items from the Learning/Value and Interest/Dynamic factors. Students were also asked to select the items which were "most important" in describing the course or instructor. Items from the Interest/Dynamic and Learning/Value factors were selected most often. In general there was a strong correspondence between an item's ability to discriminate between Best and Worst Teachers and its selection as being most important. Based upon criteria discussed, a student evaluation instrument is recommended and its development is described. The rating scale is enclosed. (Author/EVH)