BACKGROUND: Neonatal sepsis is traditionally classified as early-onset sepsis (EOS) and late-onset sepsis (LOS) disease categories. This paradigm was based on observed epidemiological data from high income settings. However, increasing availability of microbiology results from diverse settings challenges these assumptions, necessitating re-examination of neonatal sepsis classifications. OBJECTIVES: To review the literature describing the aetiology of EOS and LOS in hospitalized neonates with stratification of pathogen spectrum by low- (LIC), middle- (MIC) and high-income (HIC) country settings, to critically re-examine the continued appropriateness of the 'EOS vs. LOS' sepsis paradigm in all settings. SOURCES: PubMed was searched for peer-reviewed English full-text articles published from inception up until 8 August 2022. CONTENT: Studies often report on either EOS or LOS, rather than both. We identified only 49 original articles reporting on pathogen distribution of both EOS and LOS in the same hospital setting. Clear differences in sepsis aetiology were shown between LIC, MIC and HIC settings, with increasing importance of Klebsiella pneumoniae and decreasing importance of Group B Streptococcus in the first 72 hours of life in LIC and MIC. IMPLICATIONS: The concept of 'EOS vs. LOS' may be less useful for predicting the pathogen spectrum of neonatal sepsis in LIC and MIC, but the paradigm has shaped reporting of neonatal sepsis, and our understanding. Future neonatal sepsis reporting should utilize strengthening the reporting of observational studies in epidemiology for newborn infection (STROBE-NI) reporting guidelines and clearly describe timing of infection by day, and variation in pathogen spectrum across the neonatal period. Data identified in this review challenge the generalizability of the prevailing EOS/LOS paradigm in LIC and MIC.