This study was designed to investigate the effects of strength of descriptors in teaching evaluation items on global ratings of the instructor and the course. Six teaching evaluation forms were selected for this study with items that varied in either the descriptor in the item stem or the descriptors labeling the scale points. The forms were distributed in nine university classes from two different colleges. The total number of usable returns was 586. Analysis of variance results indicated that the version of the evaluation form has a significant effect on both instructor ratings and course ratings. These findings suggest that one should exercise caution in comparing ratings of one group of instructors with another without considering how the item stems and response options are worded.