UNLABELLED: Late onset neonatal sepsis (LOS) has a high mortality and the optimal management is poorly defined. We aimed to evaluate new expert panel-derived criteria to define LOS and characterize the current management and antibiotic susceptibility of LOS-causing organisms in Europe. A prospective observational study enrolled infants aged 4 to 90 days in five European countries. Clinical and laboratory findings as well as empiric treatment were recorded and patients were followed until the end of antibiotic therapy. Failure was defined as a change of primary antibiotic, no resolution of clinical signs, appearance of new signs/pathogens or death. Antibiotic therapy was considered appropriate if the organism was susceptible to at least one empiric antibiotic. 113 infants (median age 14 days, 62 % ≤1500 g) were recruited; 61 % were culture proven cases (28 CoNS, 24 Enterobacteriaceae, 11 other Gram-positives and 6 Gram-negative non-fermentative organisms). The predictive value of the expert-panel criteria to identify patients with a culture proven LOS was 61 % (95 % CI 52 % to 70 %). Around one third of Enterobacteriaceae were resistant to ampicillin + or cefotaxime + gentamicin but only 10 % to meropenem. Empiric treatment contained a total of 43 different antibiotic regimens. All-cause mortality was 8 % with an additional 45 % classified as failure of empiric therapy, mainly due to change of primary antibiotics (42/60).CONCLUSIONS: The expert panel-derived diagnostic criteria performed well identifying a high rate of culture proven sepsis. Current management of LOS in Europe is extremely variable suggesting an urgent need of evidence-based guidelines.