Introduction Neonatal sepsis is a leading cause of infant mortality worldwide with non-specific and varied presentation. We aimed to catalogue the current definitions of neonatal sepsis in published randomised controlled trials (RCTs). Method A systematic search of the Embase and Cochrane databases was performed for RCTs which explicitly stated a definition for neonatal sepsis. Definitions were sub-divided into five primary criteria for infection (culture, laboratory findings, clinical signs, radiological evidence and risk factors) and stratified by qualifiers (early/late-onset and likelihood of sepsis). Results Of 668 papers screened, 80 RCTs were included and 128 individual definitions identified. The single most common definition was neonatal sepsis defined by blood culture alone (n = 35), followed by culture and clinical signs (n = 29), and then laboratory tests/clinical signs (n = 25). Blood culture featured in 83 definitions, laboratory testing featured in 48 definitions while clinical signs and radiology featured in 80 and 8 definitions, respectively. Discussion A diverse range of definitions of neonatal sepsis are used and based on microbiological culture, laboratory tests and clinical signs in contrast to adult and paediatric sepsis which use organ dysfunction. An international consensus-based definition of neonatal sepsis could allow meta-analysis and translate results to improve outcomes.