A persistent problem in the evaluation of field-based projects is the lack of no-treatment comparison groups. Frequently, potential comparison groups are confounded by socioeconomic, racial, or other factors. Among the possible methods for dealing with this problem are various matching procedures, but they are cumbersome to use with multiple matching variables. Matching with a large number of variables is possible through the use of cluster analytic techniques. Specifically, principal components analysis followed by the formation of a similarity matrix based on a Euclidean distance metric may be a useful method for identifying potential comparison groups. This technique is illustrated through an example from the evaluation of the PEGASUS-PACE reading management program in the Mobile County (AL) Public School System. (Author/BW)