This study examines the friendship patterns of students in a year 7 to 10 A.C.T. high school using Vertical Integration as a grouping policy. The origins of this study can be found in the school's policy developed in 1976 and favouring vertical integration on the basis of perceived social benefits. In the development of teaching strategies at Melba High, Vertical Integration has been used with all year groups in all subject areas. Over the last three years teachers have questioned the validity of this policy. This study arose out of the demand by staff for more detailed understanding of group formation processes generally and of the link between group formation and school grouping policies at Melba High in particular. The approach used to generate data was an action research design that could feed information to the school decision-making processes. The key findings of the study were that; (a) whatever form of Vertical Integration was used, students still worked in age and sex groups by choice. (b) students tended to form friendships on a subschool basis. Recommendations resulting from these findings have been implemented in the present planning of Melba High School. The format for Vertical Integration (the method of combining years 7,8,9 and 10 in classes) has been modified to combine adjacent year groups only. (viz. years 7 and 8 as one group and years 9 and 10 as another).