The hospitalization and mortality rates of patients gradually increase following the onset and progression of liver cirrhosis (LC). We aimed to help define clinical stage and better target interventions by detecting the expression of specific metabolites in patients with different stages of LC via Q Exactive hybrid quadrupole orbitrap mass spectrometry (UPLC-Q-Exactive) technology. This noninterventional observation case-control study involved 139 patients with LC or acute-on-chronic liver failure (ACLF) in a Chinese hospital between October 2022 and April 2023. Serum specimens were analyzed for multiple metabolite levels using UPLC-Q-Exactive. Data were processed to screen for differentially accumulated metabolites (DAMs). Short time-series expression miner (STEM) analysis and enrichment analysis were performed to assess cirrhosis progression biomarkers. Following univariate and multivariate analyses, a Venn diagram indicated nine significant DAMs in common among groups. STEM analysis showed 8'-hydroxyabscisic acid, HDCA, pyruvate-3-phosphate, indospicine, eplerenone, and DEHP as significant; their levels first peaked [Child-Turcotte-Pugh (CTP) class B peaked] and then decreased with CTP grade aggravation. Significant differences among 8'-hydroxyabscisic acid, eplerenone, and DEHP were observed among LC comorbidities and between subgroups. Therefore, serum levels of six DAMs may characterize metabolomic changes, determine the severity of LC, and predict the development of ACLF.