Objective: To evaluate the implementation of switch from intravenous-to-oral antibiotic therapy with amoxicillin in neonates with early-onset infection (EOI). Design, setting and patients: A population-based multicentre cohort study. All term-born neonates with EOI were prospectively included between 1 December 2018 to 30 November 2020. Intervention: Intravenous-to-oral switch antibiotic therapy in clinically stable neonates. Main outcome measures: The primary outcome was readmission due to infection. Secondary outcomes were days of hospitalisation and antibiotic use in the pre-implementation versus post implementation period. Results: During 2 years, 835 neonates commenced antibiotics for EOI (1.5% (95% CI 1.4% to 1.6%)) of all term live births). Of those, 554 (66%) underwent a full course of treatment. There were 23 episodes of culture-proven infection (0.42 per 1000 term live births (95% CI 0.27 to 0.63)). A total of 478 of 531 (90%) neonates with probable infection underwent switch therapy. None was readmitted due to infection. The median duration of hospitalisation was 3.0 days (IQR 2.5-3.5) and 7.4 days (IQR 7.0-7.5) in the switch and intravenous therapy groups, respectively. According to antibiotic surveillance data, 1.2% underwent a full course of treatment following implementation of oral switch therapy (2019-2020), compared with 1.2% before (2017-2018). Conclusion: In clinical practice, switch therapy was safe and used in 9 of 10 neonates with probable EOI. Knowledge of the safety of antibiotic de-escalation is important as home-based oral therapy ameliorates the treatment burden for neonates, caregivers and healthcare systems. Despite the ease of oral administration, implementation of switch therapy did not increase the overall use of antibiotics.