The purpose was to compare the effects of three physical education settings (structured contact, nonstructured contact, and no contact) on attitudes of children toward peers with severe mental retardation who used wheelchairs. Contact theory (Allport, 1954) guided the study. Participants were 131 Grade 4 students (62 females, 69 males) in three intact classes that were randomly assigned to treatments. During the experimental period (4 weeks, 20 sessions, each 25 min), two children in wheelchairs were integrated into each contact class, and a special helper model was implemented. The experimental design was pretest-posttest randomized groups. Attitudes were assessed by an adjective checklist and an intention survey. A three-way ANOVA (Gender X Group X Time) revealed that females had significantly better attitudes than males. Subsequent analysis revealed that males in the structured contact group improved significantly on the adjective checklist, whereas males in the nonstructured contact group improved significantly on the intention survey.