The onset and progression of chronic liver disease (CLD) is a multistage process spanning years or several decades. Some bile acid (BA) features are identified as indicators for CLD progression. However, BAs are highly influenced by various factors and are stage- and/or population- specific. Emerging evidences demonstrated the association of structure of conjugated BAs and CLD progression. Here, we aimed to investigate the alteration of conjugated BAs and identify new features for CLD progression.Based on liquid chromatography-mass spectroscopy platform, 15 BAs were quantified in 1883 participants including healthy controls and CLD patients (NAFL, NASH, fibrosis, cirrhosis, and 3 types of liver cancer). Logistic regression was used to construct diagnostic models. Model performances were evaluated in discovery and test sets by area under the ROC curve (auROC), sensitivity, specificity, accuracy, and kappa index.Five BA glycine:taurine ratios were calculated and GCA/TCA, GDCA/TDCA, and GCDCA/TCDCA, were identified as candidates. Three diagnostic models were constructed for the differentiation of healthy control and early CLD (NAFL+NASH), early and advanced CLD (fibrosis+cirrhosis+liver cancer), and NAFL and NASH respectively. The auROCs of the models ranged from 0.91 to 0.97. The addition of age and gender improved model performances further. The alterations of the candidates and the performances of the diagnostic models were successfully validated by independent test sets (n=291).Our findings revealed stage-specific BA perturbation patterns and provided new biomarkers and tools for the monitoring of liver disease progression.This article is protected by copyright. All rights reserved.