Objective To investigate the accuracy of liver stiffness (LS) as a noninvasive index in predicting hepatic venous pressure gradient (HVPG) in patients with decompensated liver cirrhosis and the value of LS in the diagnosis of decompensated liver cirrhosis. Methods A retrospective analysis was performed for the clinical data of 88 patients with decompensated cirrhosis due to viral hepatitis or decompensated alcoholic cirrhosis who received both HVPG measurement and LS measurement by acoustic radiation force impulse (ARFI) in Department of Gastroenterology, Nanjing Drum Tower Hospital, from April 2013 to June 2021, and according to HVPG, the patients were divided into serious portal hypertension (SPH) (HVPG≥20 mmHg) group with 24 patients and non-SPH (HVPG < 20 mmHg) group with 64 patients. The two groups were compared in terms of LS, spleen stiffness, portal vein velocity, and related biochemical parameters. The t -test or the Mann-Whitney U rank sum test was used for comparison of continuous data between two groups, and the chi-square test was used for comparison of categorical data between two groups. A Pearson correlation analysis was used to investigate the correlation of different noninvasive indices with HVPG, and a Logistic regression analysis was used to investigate the association of different noninvasive indices with the risk of SPH. Receiver operating characteristic (ROC) curves were plotted for different noninvasive indices in predicting HVPG≥20 mmHg, and the area under the ROC curve (AUC), sensitivity, specificity, maximum Youden index, and corresponding cut-off value were calculated to investigate the value of each index in predicting SPH. Results Among the 88 patients, 76 had decompensated cirrhosis due to viral hepatitis and 12 had decompensated alcoholic cirrhosis. There were no significant differences between the SPH group and the non-SPH group in age, sex, white blood cell count, hemoglobin, platelet count, prothrombin time, alanine aminotransferase, aspartate aminotransferase, albumin, serum sodium, creatinine, Child-Pugh class, and spleen stiffness, while there was a significant difference in LS between the two groups ( t =-3.970, P < 0.01). The correlation analysis showed that HVPG was positively correlated with LS ( r =0.458, P < 0.001). The Logistic regression analysis showed that LS was a risk factor for SPH (odds ratio=3.941, 95% confidence interval: 1.245-12.476, P =0.020). The ROC curve analysis showed that LS had an AUC of 0.751 in predicting the onset of SPH, with a sensitivity of 54.17% and a specificity of 90.63% at the optimal cut-off value of 2.295 m/s. Conclusion In patients with decompensated cirrhosis, LS measured by ARFI is correlated with HVPG and has a certain value in the non-invasive diagnosis of decompensated cirrhosis with HVPG≥20 mmHg.