To the Editor In many neonatal units around the world, C-reactive protein (CRP) is standardly used as part of the diagnosis of neonatal sepsis, most commonly in combination with full blood cell count and clinical signs. Its negative predictive value is well established in clinical practice as part of antimicrobial stewardship to optimize antibiotic use. The meta-analysis of CRP in late-onset sepsis by Brown et al1 confirms that CRP cannot be used as a single marker of neonatal sepsis or magic bullet diagnostic test, and this is reflected in clinical practice. Currently, there is no ideal diagnostic test for sepsis, irrespective of patient age. Further, this article highlighted the continued use of a definition of neonatal sepsis based on microbiologically confirmed infection rather than sepsis. This is in clear contrast to the Sepsis-3 consensus2 definition in adults and children that instead uses the presence of organ dysfunction without requirement for microbiologic confirmation.2 This definition thus acknowledges the lack of a definitive sepsis biomarker and the significant limitations of traditional microbiologic culture, in neonates and preterm infants in particular, with typically small available blood volumes and (maternal) antibiotic exposure.