Sexual violence in armed conflict varies considerably. Within the same conflict, one group may engage in violations while another does not. Cohen argues that armed groups with low social cohesion use gang rape, especially, for combatant socialization that breaks ties to home while increasing social cohesion of the unit. Using abduction by insurgents and press-ganging by state actors as proxies for low social cohesion, she tests her argument on a cross-national data set of civil wars from 1980 to 2012. In three excellent case studies, she finds evidence for an opportunity/greed argument as well as the social-cohesion argument. She also evaluates revenge arguments for Timore-Lest and finds some evidence that groups that want to control rape can generally do so. The FMLN in El Salvador and the Falintil in Timor-Leste, both based on voluntary recruitment, limited civilian violations. However, unit commanders in Sierra Leone were reported participating in gang rape (usually going first) while recollections of gang rape by Salvadoran state agents noted that unit commanders were not present. The statistical analysis provides significant evidence that abduction for recruitment is associated with higher levels of rape by insurgents and robust significant evidence that press-ganging is associated with higher levels of rape committed by state actors, as well as robust significant evidence that the magnitude of state failure and funding through contraband is associated with higher levels of rape committed by insurgents but (for the magnitude variable) not by state actors. These findings support an opportunity/greed mechanism for rape in war. Equally important are null findings for ethnic war, gender inequality, and state troop quality. Future research needs to address these findings.