BackgroundAlthough there is evidence that family violence increased in the United States during the COVID-19 pandemic, few studies have characterized longitudinal trends in family violence across the course of initial stay-at-home orders. ObjectiveThe purpose of the present study is to investigate patterns and predictors of family violence, such as child maltreatment and harsh punishment, during the first eight weeks of the pandemic after initial stay-at-home orders in North Carolina. Participants and settingParticipants included 120 families with children ages 4–11 (53% non-White, 49% female) and a primary caregiver (98% female) living in rural and suburban areas in North Carolina. Participants were recruited based on high risk of pre-pandemic family violence exposure. MethodsCaregivers completed weekly surveys during the pandemic assessing family violence, caregiver employment status, and caregiver emotion reactivity. In addition, all caregivers completed pre-pandemic surveys on family violence. ResultsMixed-effects models revealed that family violence was highest following initial stay-at-home orders and decreased linearly over time. Higher pre-pandemic child violence exposure and caregiver unemployment were associated with higher initial family violence. Higher caregiver emotion reactivity was associated with changes in family violence across time. ConclusionsWe observed high levels of family violence following stay-at-home orders, especially in families with higher baseline violence, higher caregiver emotion reactivity, and caregiver unemployment or underemployment. These associations suggest that vulnerable families may respond to the additional stressor of stay-at-home orders with increased violence and thus need additional support in moments of crisis.