This paper considers the distinctive issues demographers face when they must forecast enrollments in a context of court-ordered desegregation. Specifically, it examines whether magnet schools strengthen a district's overall attractiveness to enrollees from outside, or whether they only siphon students away from other nonmagnet schools within the district, achieving no net effect districtwide. This issue can be clarified by examining patterns of change in grade progression rates as magnet schools are phased in. Methodology involved the analysis of student enrollment and retention for the years 1986-90 in the Kansas City, Missouri School District (KCMSD), a large urban district being transformed by the implementation of a comprehensive magnet school plan. Findings indicate that the magnet schools had enhanced the district's overall retention at most grades but did not produce actual increases at most grades, having merely slowed the overall weakening of districtwide retention. The data provide a "reality check" on the necessary judgment for crafting appropriate forecasting assumptions. Six tables and two figures are included. (LMI)