This study tests the hypothesis that disadvantaged children learn more from interaction with advantaged children in Head Start classrooms than when grouped solely with other disadvantaged children. Subjects were 32 disadvantaged children who were assigned to two experimental groups of eight each and a control group of 16. Eight advantaged children were added to each of the experimental groups. Teachers assigned to all three classes were similar in experience, teaching style and demographic characteristics. Variables of chief interest in this study were cognition, language, and socialization. Pre- and posttests on a variety of standardized and specially developed instruments evaluated the performance of all children. Videotapes were used to assess socio-emotional behaviors such as aggression and dependency. When comparisons were made between control and experimental groups results indicated support for increased educability in classes holding a higher percentage of advantaged peers. Consistent gains were evident although significance at the .05 level was not reached. The experimental children gained in task persistence and verbal skills; aggressive and dependent behaviors decreased; and self concepts improved. (WY)