Rather than examining networks of individuals as prior research has done, this study systematically examines the structure and composition of co-offending among types of criminal enterprise groups. Using social network analysis, the authors show that different types of crime groups tend to have unique co-offending patterns as measured by network composition and structure. The results also support the countercurrent of criminologists who suggest that ethnically derived categories may be misleading when trying to classify criminal enterprise groups. (Published Abstract)