Sepsis remains a leading cause of neonatal mortality, particularly in low- and lower-middle-income countries (LLMIC). In the context of rising antimicrobial resistance, the etiology of neonatal sepsis is evolving, potentially making currently-recommended empirical treatment guidelines less effective. We performed a systematic review and meta-analysis to evaluate the contemporary bacterial pathogens responsible for early-onset sepsis (EOS) and late-onset neonatal sepsis (LOS) to ascertain if historical classifications-that guide empirical therapy recommendations based on assumptions around causative pathogens-may be outdated. We analyzed 48 articles incorporating 757,427 blood and cerebrospinal fluid samples collected from 311,359 neonates across 25 countries, to evaluate 4347 significant bacteria in a random-effects meta-analysis. This revealed gram-negative bacteria were now the predominant cause of both EOS (53%, 2301/4347) and LOS (71%, 2765/3894) globally. In LLMICs, the predominant cause of EOS was Klebsiella spp. (31.7%, 95% CI: 24.1-39.7%) followed by Staphylococcus aureus (17.5%, 95% CI: 8.5 to 28.4%), in marked contrast to the Streptococcus agalactiae burden seen in high-income healthcare settings. Our results reveal clear evidence that the current definitions of EOS and LOS sepsis are outdated, particularly in LLMICs. These outdated definitions may be guiding inappropriate empirical antibiotic prescribing that inadequately covers the causative pathogens responsible for neonatal sepsis globally. Harmonizing sepsis definitions across neonates, children and adults will enable a more acurate comparison of the epidemiology of sepsis in each age group and will enhance knowledge regarding the true morbidity and mortality burden of neonatal sepsis.