Severe bacterial infections have a higher incidence in the neonatal period than at any other pediatric age. Incidence is even higher in premature babies than in term newborns, and severity is increased in the absence of early diagnosis and treatment. By contrast, clinical signs are nonspecific and sometimes trivial, and biomarkers perform poorly during the first 24 hours of infection. For decades, this has led to having too many children treated for extended periods with broad-spectrum antibiotics. Today, the challenge is to prescribe antibiotics in a targeted way, by identifying truly infected newborns. Over the last ten years, major paradigm shifts have occurred and should be taken into account, as a result of growing awareness of the ecological impact of early antibiotic therapy, notably antibiotic resistance, by choosing the narrowest spectrum antibiotic and stopping antibiotic therapy as soon as the diagnosis of infection has been reasonably ruled out. Among the biological tests, the most important are blood cultures. At least one blood culture, taken under aseptic conditions, of sufficient volume (1 to 2 mL), and using pediatric bottles must be taken as soon as the decision to treat has been made, before starting any antibiotic therapy. The bacteria responsible for early-onset bacterial neonatal infections (EBNI) have not changed significantly over recent years and remain dominated by Group B Streptococcus and Escherichia coli, which are the main targets of treatment. GBS is largely predominant in full-term infants, but the proportion of infections due to E. coli increases with prematurity.