This study was designed to begin testing the proposition that returning or serving mildly handicapped students in the general classroom may jeopardize their academic self-perceptions and to assess students' preferences regarding extra academic assistance. Participants were regular (N = 287) and special education (N = 175) students in three midwestern cities. The students completed the Children's Self-Perception Profile for Learning Disabled Students (SPPLDS) and accompanying social comparison measure (Renick & Harter, 1988), as well as a measure of locus of control consisting of the cognitive section from Connell's (1985) Multidimensional Measure of Children's Perceptions of Control. Students also were asked to choose from whom and where they would prefer to receive extra academic assistance. Data on each student's educational handicapping condition, level of special education programming, curriculum modification, and extent of team teaching utilized were provided by classroom teachers;Initial analyses by district indicated differences in self-perceptions; for all academic self-perception areas regular education students in District A scored the same as the special education students (combined for all three districts) and lower than those in Districts B and C. Special education students rated Powerful Other (F (2,436) = 14.67, p <.001) and Unknown control (F (2,436) = 13.22, p <.001) as more important in obtaining grades than did the regular education students. Significant differences between regular education and special education students' choices were identified on all nine social comparison items. Regular education students chose other regular education students for comparisons 94%; special education students chose regular education students for social comparison 65% for the time for academic areas and 65% for nonacademic areas. Choice of social comparison was also significant by level of special education program; however, SPPLDS domain scores did not vary by level. Students overwhelmingly chose to leave the regular classroom for academic assistance; choice of provider varied by group. Students were more apt to chose the teacher with whom they currently spent the most time;These results endorse cautious piloting of mainstreaming or team taught models; students' academic self-perceptions varied by ability but not by setting. However, more input from students regarding their preferences and reasons for choices should be included in future integration projects.