Some of the theoretical and methodological problems with current practices in evaluating instruction at the higher education level are reviewed. Controversy over the evaluation of instruction in higher education has resulted at least in part from inadequate instrumentation. The instruments for instructional rating now used include administrator evaluations, colleague evaluations, self-evaluations, and student evaluations. It is suggested that the teaching-learning relationship must be the focus of meaningful evaluations of instruction. A multidimensional scaling technique for precisely evaluating instruction that is based on a cognitive perspective of student learning is presented. The technique provides a means of assessing cognitive processes in a manner that has been found to be precise and reliable and enables the direct assessment of teaching effectiveness. A major advantage is that the proposed system does not require advance knowledge of all relevant criteria on which student responses are based, although specific attributes used for making distance judgments can be interpreted from the results. Making complete paired comparisons on all concepts that are provided allows multiple criteria to be simultaneously examined and reduces the influence of social desirability factors. This evaluation system also establishes explicit standards for direct evaluation; provides relevant information for administrators, instructors, and students; is precisely ratio-scaled; and permits powerful time-series analysis of the learning process, permitting instructors to monitor their effectiveness throughout the term. (SC)