Antibiotics are administered to the vast majority of preterm newborns and to a substantial proportion of term infants in the hours after birth due to risk for early-onset sepsis. The approaches taken to determine which newborns should be evaluated for early-onset sepsis, and what type and duration of antibiotics are administered, are important elements of neonatal antibiotic stewardship. The use of multivariate prediction models for sepsis risk assessment among infants born ≥35 weeks’ gestation can safely reduce the use of empiric antibiotic therapy. Approaches incorporating serial physical examination may also contribute to decreasing empiric antibiotic exposure among such infants. Among infants born <35 weeks' gestation, delivery characteristics can be used to identify preterm infants at low enough risk of early infection that empiric therapies are not required. Data informing the epidemiology, microbiology and antibiotic susceptibility patterns of early-onset sepsis pathogens can be used to optimize antibiotic choice for empiric and targeted antibiotic therapy to ensure that effective therapies are administered, while decreasing the risks associated with broad-spectrum antibiotic exposure. Optimal use of blood culture and time to positivity data can also contribute to decreasing the risks associated with prolonged antibiotic administration in the face of sterile cultures.