BackgroundFour decades of conflict has indelibly impacted the lives of Afghans, exposing many to different forms of trauma. The aim of this paper investigate a hypothesis that (mostly war related) trauma is a key driver of partner violence in Afghanistan.Methods1,463 women aged 18–48 were recruited into a randomised controlled trial (RCT) to evaluate a women empowerment intervention in 8 villages of Kabul and Nangarhar provinces. The women were interviewed at baseline. The analysis uses multivariable logistic regression and structural equation modelling (SEM) to describe relationships between measures.Results57.4% of women reported exposure to one of four types of trauma: 23.3% an armed attack, 39.4% had felt close to death, 10.6% witnessed a friend or family member being killed and 21.4% witnessed the death of a stranger or someone unknown. Trauma exposure was associated with being older, Pashtan, madrassa educated, and food insecure. Women who were trauma exposed were more likely to have ever experienced IPV, have hit their children in the last 4 weeks, and be hit by a sibling or relative of their husband or their mother-in-law in the last year. They held less patriarchal personal gender attitudes and perceived the community to be more patriarchal. The SEM showed that all pathways between trauma exposure and IPV were ultimately mediated by either (mostly mental) ill-health or quarrelling, but not both of these. There were multiple paths through which trauma exposure impacted women’s past year experience of physical IPV. One was mediated by childhood trauma exposure and a latent variable for ill health. Other paths were mediated by women’s education and personal gender attitudes and ill-health, or else by quarrelling. Trauma exposure was related to lower educational levels. Another path was mediated by less patriarchal personal gender attitudes and ill health. Community gender attitudes was a mediating variable on a path which was also mediated by ill health and another mediated by quarrelling. It was also a mediator on a path which included personal gender attitudes and ill-health. Food insecurity mediated another path with ill health. It was also connected to childhood trauma, community gender attitudes and educational level.ConclusionTrauma exposure due to conflict will persist until the conflict ends but the impact on women can be ameliorated. This analysis suggests interventions to reduce women’s exposure to IPV should focus on reducing poverty, changing social norms on gender, providing relationship skills to help reduce quarrelling and supporting women’s mental health.