The relationship between student/teacher attitude similarity and teacher evaluations was studied in a group of 222 students who completed a 24-item attitude scale, once for themselves and once for their instructors, and a 28-item teacher rating form (TRF). A significant, moderate-sized correlation between assumed similarity and TRF scores for students pooled across classes was found. However, the size of the relationship dropped dramatically once the relationship was examined with all possible teacher effects removed. When controlling for teachers, students who perceived their own attitudes as similar to the instructor's assigned only slightly higher TRP scores than attitudinally dissimilar students. Analysis of component scores revealed that the set of 24 self-rating variables and the set of 24 perceived rating variables explained more TRF variance than the set of 24 individual similarity scores. Only one of the two sets of component scores contributed significantly to the prediction of TRF scores independent of all other effects. The relationship between perceived teacher attitudes and TRF scores was fairly substantial, and it is concluded that the perceived scores appear to be the best predictor of ratings. However, it is impossible to determine whether student differences in their perceptions of teacher attitudes affect their ratings of teacher effectiveness or vice versa. Theories of similarity and attraction and college classroom similarity research are reviewed, and a sample attitude scale is appended. (SW)