This review of research focuses on policies and practices that result in placing students in groups that are more or less homogeneous with respect to academic performance. Recent analysis of data from the National Education Longitudinal Study of 1988 provides the largest and best-controlled multi-year study of ability grouping ever conducted (Braddock and Slavin). The outcomes of scores of studies have been similar, and these outcomes are discussed in the following categories: (1) opportunities to learn; (2) ability grouping and achievement; (3) ability grouping and segregation; (4) ability grouping and intergroup relations; (5) ability grouping, self-esteem, and feelings of inferiority; (6) ability grouping, delinquency, and dropouts; and (7) alternatives to ability grouping. Ability grouping must end because it is ineffective, harmful to many students, and damaging to interracial relations and democratic society. Effective and practical alternatives exist. Public schools must provide more equitable access to learning opportunities that develop reasoning, inference, and critical thinking skills. Major school restructuring will be necessary to develop the needed alternatives. A 71-item list of references is included. (SLD)