Studies on the criterion validity of student evaluation-of-instruction instruments are analyzed, and recommendations are offered for future research into student evaluation of instruction. The main problem, and probably the reason for the lack of validity studies, is that it is difficult to agree on what the criteria of effective teaching should be. One method of dealing with the problems of research in student evaluation-of-instruction instruments is to select a measurable definition of teaching effectiveness. Since the ultimate criterion of teaching effectiveness is student learning, there is general agreement that an appropriate and defensible criterion is the amount that students learn as measured by achievement examinations. Attention is directed to: studies incorporating achievement scores and random assignment; studies incorporating achievement scores adjusted for ability; studies incorporating achievement scores not adjusted for ability; and a meta-analysis of student ratings and student achievement. Studies using criterion measures in conjunction with achievement and studies using criterion measures other than achievement are also reviewed. Tables are presented to summarize the studies that examined the relationship of student ratings of instruction and criterion measures. Although parallel data were not reported in all the studies, the table shows the largest significant correlation reported in each study. These largest correlations are squared to indicate the proportion of variance shared by the criterion and the student ratings. The majority of the investigations reported significant positive correlations between student ratings of instruction and criterion measures of effective teaching; however, the correlation between the ratings and criteria were usually modest. A bibliography is appended. (SW)