BACKGROUND: Given the inconsistencies in guideline recommendations on optimal timing of empirical antibiotic discontinuation, this study aimed to systematically review and assess the efficacy and safety of different discontinuation strategies for suspected early-onset sepsis (EOS). METHODS: We included randomized controlled trials (RCTs) comparing different durations of empirical antibiotics for suspected neonatal EOS. The data sources included PubMed, Embase, Cochrane CENTRAL, Web of Science, CINAHL, CNKI, SinoMed, and clinical trial registration websites. Details of the discontinuation regimens and outcomes were extracted into a predefined form. RESULTS: This review included 11 unique studies evaluating nine different antibiotic discontinuation regimens. Eight RCTs compared various discontinuation regimens and had heterogeneous clinical characteristics, allowing only descriptive analyses. Three other studies found procalcitonin-guided discontinuation can safely reduce the duration of antibiotics (very low certainty) compared with clinical judgement discontinuation, without increasing the 14-day reinitiation rate (moderate certainty). Notably, there is a paucity of studies comparing guideline-recommended short courses (36-48 h) with longer courses or biomarker-guided discontinuation regimens. Of the 11 RCTs, 10 were assessed as high risk of bias, with only one at low risk. CONCLUSIONS: There is currently insufficient evidence to determine the optimal strategy for discontinuing empirical antibiotics in suspected neonatal EOS. IMPACT: This review found insufficient evidence to determine the optimal strategy for discontinuing empirical antibiotics in suspected EOS. This study highlights the need for more robust evidence to support guideline recommendations for discontinuing antibiotics in suspected EOS, particularly in resource-limited settings. Future research should focus on large, well-designed RCTs comparing short-course and biomarker-guided strategies, as well as economic analyses.