Bacterial infection is the most common complication in paediatric oncological patients during cancer treatment. A suitable tool for early prediction of unfavourable course of infection is still needed. We performed a prospective longitudinal observational study to evaluate of the role of serum biomarkers (C-reactive protein, procalcitonin, interleukin-6, presepsin) in the early diagnosis of bacteraemia (gram-negative versus gram-positive) in patients with haematological malignancies. We observed 69 febrile episodes in 33 patients (17 male, 16 female; 1.5-18.9 years, mean 7.31 years, median 5 years). Within this sample, there were 22 cases of positive blood cultures, 16 cases of sepsis, 38 cases of fever with no signs or symptoms of sepsis, and two deaths from infectious complications. All markers tested had good negative predictive value (73% -93%). CRP was characterized by good specificity for registration bacteraemia (96%, 95% CI: 85% -99%), but other results were inconclusive. We identified comparably balanced sensitivity (64% -81%) and specificity (61% -88%) for interleukin-6 and procalcitonin, and we proved their quality to predict positive blood culture and clinical signs of sepsis as well. Patients with gram-negative bacteraemia had significantly elevated levels of PCT and IL-6 in comparison with a group of patients with gram-positive bacteraemia (p = 0.04 for PCT and p = 0.005 for IL-6). Presepsin was characterized by poor specificity (27%, 95% CI: 15% -43%) and positive predictive value (24%, 95% CI: 12 -39%) for predicting bacteraemia, and by better sensitivity (84%, 95% CI: 55% -98%) and specificity (58%, 95% CI: 42% -73%) for predicting clinical signs of sepsis. Key words: C-reactive protein, procalcitonin, interleukin-6, presepsin, fever, sepsisBacterial infection is the most common treatment-related complication in patients with haematological malignancies [1]. Documented mortality associated with paediatric febrile neutropenia is 2% [2]. The potential for early diagnosis of bacteraemia through serum biomarkers has been the subject to extensive research [3]. In 2012 Phillips et al published large meta-analysis of 25 studies exploring 14 different biomarkers in 3,585 episodes of febrile neutropenia. CRP, PCT and IL-6 were subject to quantitative meta-analysis. The bivariate estimates of diagnostic precision of these biomarkers and outcomes were done. Data were available for meta-analysis for CRP for microbiologically or clinically documented infection (results: cut off > 50mg/dl, sensitivity 65%, specificity 73%), for PCT assessing microbiologically or clinically documented infection (results: cut off > 0.2 mg/ml, sensitivity 96%, specificity 85%), for IL-6 reporting microbiologically or clinically documented infection (results: cut off > 235 pg/ml, sensitivity 68%, specificity 94%), and gram-negative bacteraemia (results: cut off > 1000 pg/ml, sensitivity 78%, specificity 96%). Huge inconsistencies and heterogeneity in the studies included in this review were the most important limiting factors [...