The analysis notes that the sources of gang cohesiveness are mainly external to the group, in contrast to findings on most groups previously studied. Gang delinquency is partly the consequence of gang interaction. It also serves as an additional reinforcer of gang cohesiveness. The concepts and approaches currently available for measuring group cohesiveness are somewhat insufficient for application to gang research. The research in Los Angeles suggests a more promising approach. Data were collected over 4 years by detached workers who observed gang members. The gang workers' contact reports covered 4 black gang clusters with 16 separate gang groups. Contact reports gave accounts of all persons seen by the workers each day. The data were used to examine companionship patterns within the groups, including the number of cliques, the percentage of members in cliques, and other factors. The preliminary data indicated that the increase in cohesiveness among the younger males and the decrease in cohesiveness among the older males paralleled similar trends in recorded offense rates. The indexes of cohesiveness also appeared to be sensitive, although it is not yet clear how directly they are related to gang offense patterns. Figure, tables, and reference notes