Analysts of institutional performance have occasionally used a peer grouping approach in which they compared institutions only to other institutions with similar characteristics. Because analysts historically have used cluster analysis to define peer groups (i.e., the group of comparable institutions), the author proposes and demonstrates with actual peer grouping data a method for diagnosing the robustness of a peer grouping effort. This method measures the effect of alternative clustering approaches through the calculation of the cluster sensitivity index (CSI). This index, however, is flexible enough to help in other kinds of comparisons that do not apply cluster analysis or do not involve institutions of higher education. (Contains 8 figures.)