Analysis of covariance (ANCOVA) has been recommended as one vehicle with which to evaluate special education and other intervention impacts (M. J. Taylor and M. S. Innocenti, 1993). Common misinterpretations of this methodology for these purposes are explained. These misapplications of ANCOVA include: (1) ignoring the assumption of homogeneity of regression; (2) using ANCOVA even given a lack of random assignment of subjects to groups; and (3) lack of attention to reliability of covariate scores. It is noted that such misapplications of ANCOVA "can mistakenly make compensatory education look harmful" (D. T. Campbell and A. Erlebacher, 1975, p. 597). At the very least, researchers should exercise caution in using ANCOVA to consider differences between groups, looking carefully at conditions that may bias the data and give false information about treatments. Contains 31 references. (SLD)