BACKGROUND: Globally, about 30% of women have experienced physical or sexual violence, or both, from an intimate partner during their lifetime. Associations between poverty and women's increased risk of intimate partner violence have been observed. We therefore aimed to assess the effect of a violence prevention intervention delivered to women participating in a group-based microfinance scheme in Tanzania. METHODS: We did a cluster randomised controlled trial among women taking part in a microfinance loan scheme in Mwanza city, Tanzania. A microfinance loan group was only enrolled if at least 70% of members consented. We randomly assigned the microfinance groups in blocks of six to receive either the intervention (ie, the intervention arm) or be wait-listed for the intervention after the trial (ie, the control arm). Women in both arms of the trial met weekly for loan repayments. Only those in the intervention arm participated in the ten-session MAISHA intervention that aims to empower women and prevent intimate partner violence. Given the nature of the intervention, it was not possible to mask participants or the research team. The primary outcome was a composite of reported past-year physical or sexual intimate partner violence, or both. Secondary outcome measures were past-year physical, sexual, and emotional intimate partner violence; acceptability and tolerance of intimate partner violence; and attitudes and beliefs related to intimate partner violence. These outcomes were assessed 24 months after the intervention. An intention-to-treat analysis was done, adjusting for age, education, and baseline measure of the respective outcome. The study is registered with ClinicalTrials.gov, number NCT02592252. FINDINGS: Between September, 2014, and June, 2015, 66 (65%) of 101 microfinance groups approached in the study area met the trial eligibility criteria and were enrolled, of which 33 (n=544 women) were allocated to the intervention arm and 33 (n=505 women) to the control arm. Overall, 485 (89%) of 544 women in the intervention arm and 434 (86%) of 505 in the control arm completed the outcomes assessment. Among the intervention arm, 112 (23%) of 485 women reported past-year physical or sexual intimate partner violence, or both, compared with 119 (27%) of 434 in the control arm (adjusted odds ratio [aOR] 0·69, 95% CI 0·47-1·01; p=0·056). Women in the intervention arm were less likely to report physical intimate partner violence (aOR 0·64, 95% CI 0·41-0·99; p=0·043) and were less likely to express attitudes accepting of intimate partner violence (0·45, 0·34-0·61; p<0·0001) or beliefs that intimate partner violence is a private matter (0·51, 0·32-0·81; p=0·005) or should be tolerated (0·68, 0·45-1·01; p=0·055). There was no evidence of an effect on reported sexual or emotional intimate partner violence. There were no reports that participation in the trial had led to new episodes of violence or worsening of ongoing violence and abuse. INTERPRETATION: Reported physical or sexual intimate partner violence, or both, was reduced among women who participated in the intervention arm, although the effect was greater for physical intimate partner violence, suggesting that intimate partner violence is preventable in high-risk settings such as Tanzania. FUNDING: Anonymous donor and STRIVE Consortium.