DOI: 10.1038/s41372-018-0081-3 Asymptomatic term and late-preterm newborns with risk factors for early onset sepsis commonly undergo laboratory evaluation and receive empiric antibiotic therapy. Some have challenged the rationale for current “rule-out sepsis” practices, arguing that they lead to unnecessary overtreatment and healthcare costs. A series of recent clinical studies has explored scheduled serial observations as an alternative to laboratory testing and empiric antibiotics for asymptomatic newborns with historical risk factors for sepsis. These studies have shared the conclusion that serial observation is safe and cost-effective for well-appearing term and late-preterm babies, but they are also somewhat speculative because culture-proven early onset sepsis is an extremely low prevalence diagnosis. Here, we review the evolving consensus of optimal rule-out sepsis practices. We examine chorioamnionitis as an example of a problematic risk factor that has contributed to the controversy surrounding this topic. We also discuss how introduction of online sepsis risk calculators has allowed more precise delineation of a patient’s chances of developing culture-proven infection. Finally, we analyze existing data from published studies to estimate the number needed to harm (NNH) when an observation-based strategy is used instead of a risk-based approach. We conclude that, if harm is defined as death or serious sepsis complications such as hemodynamic instability or neurologic injury, the NNH is 1610, compared to an NNH of 7 and 2.9 for IV infiltrates and delayed breastfeeding, respectively—two common and potentially consequential complications of NICU admission for a rule-out sepsis. We believe that the differential between risk of serious harm from observing a well-appearing term or late-preterm newborn with risk factors for sepsis and the risk of less significant but common NICU complications argues in favor of the ongoing trend toward less aggressive management of newborns with sepsis risks.