The question of regression to the mean is discussed in the contexts of the Title I Evaluation and Reporting System (TIERS) and the Bilingual Education Evaluation System (BEES), both of which involve assessing the achievement growth of project students from pre- to post-test. The correction formula recommended for use with TIERS was designed to eliminate regression between selection and pre-test; for BEES, the idea of a separate selection test was eliminated and an attempt was made to eliminate regression between the selection/pre-test and the post-test and to estimate total group correlation from the actual correlation in the selected sub-group. How much regression there is to correct for and how well alternative correction procedures work are discussed. (SLD)