Four methods of deriving indicators of school system performance were investigated. All four methods involve the regression of a measure of student performance at one grade level on student performance measures at an earlier grade level. Deviations of school system means from the regression surface are used as indicators of school system performance. The four methods were: (1) matched-longitudinal data using individual student scores for the regression analyses, (2) matched-longitudinal data using system means for the regression analyses, (3) unmatched-longitudinal data using system means, and (4) cross-sectional data using system means. The first two methods yielded highly similar results; however, the results of the other two methods were quite dissimilar from methods one and two and from each other. It was concluded that the latter two methods are not reasonable substitutes for the first two methods. The stability of the results of the first two methods was investigated by comparing results for one subsample of students to results for a second subsample of students. The deviations of the school system means from the regression surface for one subsample were highly correlated with the corresponding deviations for the second subsample.