Introduction: Although poverty is often viewed as a driver of intimate partner violence (IPV), few studies have examined how changes in poverty influence men’s IPV perpetration over time. One sensitive marker of poverty is food security, or the ability of households to obtain adequate nutrition. The influence of food security has been associated with IPV in cross-sectional studies, but here we examine whether a causal relationship exists.Methods: We followed a cohort of 2,479 men living in a peri-urban settlement near Johannesburg, South Africa during the period of February 2016-August 2018. Using self-completed audio-assisted questionnaires, men responded at baseline (T0), 12 months later (T1), and 24 months later (T2). Questions about current food security, household type, childhood conditions, and past-year IPV perpetration were asked. We used cross-lagged dynamic panel modeling to assess whether food insecurity changes influenced men’s IPV perpetration over time.Results: A large proportion of men (46.9%) reported perpetrating IPV at baseline, but this proportion reduced over time (39.2% and 26.5% reported IPV at midline and endline, respectively). Less than half (41.4%) of participants reported food insecurity at T0, and this stayed consistent at T1 and T2. In cross-sectional bivariate analysis, food insecurity was higher among those men perpetrating IPV than among those who used no violence at all three time-points. In a cross-lagged dynamic panel model, food insecurity predicted men’s use of IPV one year later. Food insecurity was associated with a small, but significant, longitudinal impact on men’s IPV use (standardized coefficient=0.08, p=0.035) and the model fit indices were strong (RMSEA=0.027, CFI=0.993). This longitudinal association between food insecurity and IPV perpetration persisted in a final model controlling for housing status, age at baseline, and childhood exposure to abuse (coef=0.10, p=0.022) that had strong fit (RMSEA=0.015, CFI=0.998). In a separate model, IPV perpetration did not predict later food security, suggesting a causal relationship from poverty to IPV perpetration.Conclusions: Food insecurity and housing, as distinct markers of poverty, had small but persistent longitudinal effects on men’s perpetration of IPV in a peri-urban settlement in South Africa. Addressing IPV perpetration will require examination of broader structural challenges, such as livelihoods and food security, particularly in settings with endemic poverty. Future interventions should consider livelihood strategies to improve health outcomes for both women and men.