Intimate partner violence is an important public health problem, with far-reaching consequences for women's physical and emotional health and social well-being. There is evidence that intimate partner violence is preventable. The MAISHA study, a randomised controlled trial of the impact of a gender training intervention on intimate partner violence for women in Tanzania, found that those who participated in gender training were less likely to report past-year physical violence. As part of the study, a sample of women participated in longitudinal qualitative enquiry. To better understand the processes of change associated with intimate partner violence prevention, we explored narratives from in-depth interviews and focus group discussions with women who had participated in the training. The study drew on feminist political theory on agency and change, which we sought to understand in a setting with high rates of intimate partner violence that a standalone intervention was unlikely to change. The study found that gender training, which seeks to develop political consciousness and transformation, can promote a sense of efficacy amongst participants who feel validated through the collective learning process. Important yet under examined cognitive elements of change processes deserve more attention in the design, delivery and evaluation of violence prevention gender training.