Background Despite significant investment and global improvements in neonatal mortality rates over the past three decades, neonatal mortality in Zimbabwe, like in other sub-Saharan countries, has continued to rise. Ensuring access to institutional delivery is an important factor in improving both neonatal and maternal outcomes. MethodsThis study uses data from the Sanitation Hygiene Infant Nutrition Efficacy (SHINE) trial, a cluster-randomized community-based trial, to examine the risk factors, birth practices and infant outcomes among women having institutional or non-institutional deliveries in rural Zimbabwe. Results 11.9% (n=536) fetuses in the study were delivered by 529 mothers outside a health institution. These women were more likely to have a lower socioeconomic status than those who delivered in a health institution. Additionally, hygiene practices were poorer and interventions to minimise neonatal hypothermia less commonly utilised outside of health institutions. Infants with non-institutional deliveries were born on average 1.15 weeks (95%CI 0.84-1.46) earlier than those who were delivered in a health institution; their risk of being born preterm was more than twice as high (RR 2.20, 95%CI 1.92-2.53). The risk of stillbirth and neonatal death among infants born outside a health institution was more than two-fold higher (RR 2.38, 95%CI 1.36-4.15; and 2.01, 95%CI 1.31-3.10, respectively) that among infants born in a health institution. ConclusionsOur findings indicate that premature onset of labor, rather than maternal choice, may be the reason for many non-institutional deliveries in low-resource settings. This most often occurs amongst women from lower socioeconomic backgrounds who cannot reach a health institution quickly, and so are attended by untrained caregivers in less hygienic conditions. This cascade of events results in a two-fold higher risk of stillbirth and neonatal mortality amongst children born outside of a health institution. Interventions for primary prevention of preterm delivery will be crucial in reducing neonatal mortality in Zimbabwe. Trial registration The trial is registered with ClinicalTrials.gov, number NCT01824940.