This presentation uses data from a long-term juvenile diversion counseling program to illustrate a sample method of estimating program cost. Several problems in using average cost figures are first presented. Then a method is described in which discriminant analysis can be used to refine cost comparisons. This makes it possible to produce meaningful comparisons between programs which differ substantially overall, but serve a somewhat overlapping clientele. Problems inherent in the discriminant analysis method are also mentioned. (Author/BP)