This synthesis extends part of Peter Cohen's well-known examination of the association between student ratings of instruction and student achievement. The results of increasing the number of specific instructional dimensions considered and of preserving more of the information in the evaluations by multiple coding into these dimensions are compared with Cohen's findings. Extending the analysis further by changing the unit of analysis and the method of averaging correlations generally decreases the size of the associations. The instructor's preparation and organization, clarity and understandableness, stimulation of interest, motivation of students to reach high standards, encouragement of discussion and openness to others' opinions, and elocutionary skills consistently explain 10% or more of the variance in student achievement regardless of the method of averaging or the unit of analysis used. The size of the association between an instructional characteristic and student achievement, as one indicator of the characteristic's importance, is compared with three other indicators (extent to which students say the characteristic is important, extent to which faculty say it is important, and size of its correlation with the overall evaluation of the instructor).