Objective To establish a modified sequential method based on clinical and laboratory indicators for assessment of whether febrile neonates were complicated with infection and evaluate its application value. Methods The clinical data of 230 febrile neonates admitted to the NICU from October 2015 to October 2017 were retrospectively analyzed. A modified sequential method based on clinical and laboratory indicators was established to evaluate whether febrile neonates were infected or not, and to compare it with the laboratory scoring method. Results In the 230 full-term febrile neonates, 112 (48.7%) were diagnosed with sepsis. The general condition, age, urinary routine, procalcitonin, C-reactive protein and neutrophil absolute count of children with sepsis were evaluated and classified according to the modified sequential method. The incidence of sepsis neonates was significantly higher than that of non-sepsis neonates. The sensitivity, negative predictive value and negative likelihood ratio of modified sequential method in the diagnosis of sepsis were 98.2%, 94.8% and 0.02, and the laboratory score was 66.1%, 72.9% and 0.40, respectively. Two children with sepsis were misjudged as non-sepsis by modified sequential method, and 38 children with sepsis were misjudged as non-sepsis by laboratory scoring method. Conclusions Modified sequential method can identify infected and non-infected children. It has high sensitivity for infected newborns and higher application value than laboratory score.