Background and aim Both breastfeeding and the use of human milk are strategies that provide better conformation to health throughout an individual’s life and bring countless short- and long- term benefits, which are well established in the scientific literature. For at-risk newborns (NBs), these strategies are crucial interventions to enable neonatal survival with better quality of life due to the distinctive and complex composition of human milk, which serves as personalized food-medicine-protection. However, there is limited knowledge about breastfeeding practices in high-risk NBs. The aim was to estimate the duration of EBF and to investigate the effect of risk at birth on EBF discontinuity in the first six months of life’. Methods This cohort study included 1,003 NBs from a high-risk referral center, followed up from birth to the sixth month of life, between 2017 and 2018. Correspondence and cluster analysis was used to identify neonatal risk clusters as the main exposure. The object of interest was the time until EBF discontinuity. The Kaplan-Meier methods and the Cox proportional hazards model were used to estimate the hazard ratio and 95% confidence intervals. Results The prevalence and median duration of EBF decreased proportionally in the three groups. The multiple model revealed a gradient in EBF discontinuity, which was 40% higher in risk group 1 and 111% higher in risk group 2 compared to healthy full-term NBs. Additionally, EBF during hospitalization predicted a longer median duration of this practice for high-risk NBs. Conclusion This study confirms a high proportion of high-risk NBs who have EBF discontinued before six months of life. The risk of EBF discontinuity is higher in risk groups, with a gradual effect even when adjusted by several factors. Effective interventions are needed to promote, protect, and support breastfeeding in different profiles of risk-at-birth groups.