Objective:To establish a noninvasive prediction model for liver fibrosis in chronic hepatitis B (CHB) virus infection patients with alanine aminotransferase (ALT) less than upper limit of normal level.Methods:A total of 183 CHB patients with ALT <40 U/L admitted in the First Affiliated of Wenzhou Medical University from January 2014 to September 2017 were enrolled. There were 149 cases of non-marked liver fibrosis (F0-F2) and 34 cases of marked liver fibrosis (F3-F6) according to the Ishak scoring system. The clinical data, liver stiffness measurement (LSM), liver ultrasound imaging, serum biochemical features and hepatitis B virus related indexes of patients were retrospectively analyzed. The independent predictors of liver fibrosis were screened and a prediction model was constructed based on the results of multivariate logistic regression analysis. The receiver operating characteristic (ROC) curve was applied to evaluate the established model and other indicators in predicting liver fibrosis.Results:Multivariate logistic regression analysis showed that LSM ( OR=1.327, 95% CI 1.149-1.531), liver ultrasound scores ( OR=6.610, 95% CI 2.704-16.156) were independent predictors of liver fibrosis. The area under ROC curve(AUC)of the established model (LU) in predicting liver fibrosis was 0.873(95% CI 0.799-0.947), which was significantly greater than that of the ultrasound score, LSM, FIB-4, APRI and GPR (AUC=0.790, 0.804, 0.654, 0.673 and 0.770; Z=3.394, 1.982, 2.077, 3.168 and 2.165, all P<0.05 or <0.01). With the cut-off value of -1.787, the sensitivity and specificity of LU model were 85.3% and 77.2%, respectively. Conclusion:The established model (LU) can effectively predict the liver fibrosis in CHB patients with ALT less than upper limit of normal.