We use a randomized experiment in Kenya to study the impact of unconditional cash transfers on intimate partner violence. Transfers to women of on average USD 709 led to a 0.26 standard deviation (SD) decrease in physical violence, and transfers to men to a 0.18 SD decrease. Sexual violence was reduced after transfers to women (0.22 SD), but not men. We construct a theory which together with our empirical findings suggests that husbands use violence to extract resources, but dislike it otherwise. We observe large and significant spillovers: nonrecipient women in treatment villages report a 0.16 SD reduction in physical violence.