Introduction:The aim of research was to assess the melatonin concentrations in the early neonatal period as a predictor of adverse outcomes of late neonatal period in preterm infants and to estimate its optimal predictive cut-off values. Materials and methods: A total of 115 preterm infants admitted to the neonatal intensive care unit were screened for eligibility, five did not meet the criteria, six parents declined the participation. So, a total of 104 preterm infants with gestational age 25-34 weeks were included in research. The concentration of melatonin in urine was determined by the Enzyme Immunoassay method (Human Melatonin Sulfate ELISA kit, Elabscience, China). The Mann-Whitney U-test and analysis of the receiver operating characteristic (ROC) curve were used in statistical analysis. Results: Analysis of the ROC curves has revealed optimal cut-off values for urinary melatonin concentration to predict late outcomes. Melatonin concentration below 3.58 ng/ml with sensitivity of 72% can predict development of retinopathy of prematurity (ROP) (AUC = 0.73; 95% confidence intervals (CI) 0.61-0.86). Good diagnostic accuracy (AUC = 0.80; 95% CI 0.67-0.93) has been shown for bronchopulmonary dysplasia (BPD). The optimal cut-off value for melatonin concentration in BPD prediction is 3.71 ng/ml (sensitivity 80%, specificity 64%). Urinary melatonin concentration below 3.79 ng/ml can be associated with late-onset sepsis (AUC = 0.76; 95% CI 0.64-0.87; sensitivity 72%; specificity 62%). There were no significant associations between melatonin concentration and necrotizing enterocolitis (P = 0.912). Conclusion: Urinary melatonin concentration below the certain cut-off values in the early neonatal period may serve as one of the predictors of adverse outcomes such as BPD, ROP, and late-onset sepsis in the late neonatal period in preterm infants.