Background: This study sought to compare the predictive value of NT-proBNP, sST2 and MMPs in HF with different ejection fractions from a population in southern China. Methods: A cross-sectional study was conducted on 113 HF patients admitted to Fujian Provincial Hospital from December 2016 to March 2018.The patients were divided into three subgroups: 60 cases in HFpEF group (LVEF≥50%), 28 cases in HFmrEF group (41% ≤ LVEF≤49%) and 25 cases in HFrEF group (LVEF≤40%). ELISA method was applied to detect the concentrations of sST2, MMP-2 and MMP-9. Electrochemical luminescence immunoassay was applied to detect the concentration of plasma NT-proBNP. Univariate and multivariate Cox and logistic regression models were used to analyze the diagnostic significance of these plasma biomarkers in HF patients. Kaplan-Meier survival curves were used to assess the prognostic value of sST2 in the incidence of long-term adverse events during study. Results: This study showed that plasma sST2 levels in HFrEF or HFmrEF patients were significantly higher than in HFpEF patients. Plasma levels of MMP-2 and MMP-9 in HFrEF patients were apparently higher than in HFpEF or HFmrEF patients. For the diagnosis of HFpEF, the AUC of NT-proBNP was higher than that of sST2, MMP-2 and MMP-9, which were 0.881, 0.717, 0.705 and 0.597, respectively. For the diagnosis of HFmrEF, the AUC of plasma sST2 was higher than that of MMP-2, MMP-9 and NT-proBNP, which were 0.799, 0.678, 0.676 and 0.793, respectively. For the diagnosis of HFrEF, the AUC of plasma NT-proBNP, sST2, MMP-2, and MMP-9 were 0.945, 0.820, 0.814, and 0.774 respectively. Spearman correlation analysis showed that plasma sST2 levels were significantly correlated with plasma MMP-2, MMP-9 and NT-proBNP levels. Further logistic regression analysis showed that except MMP-9, the biomarkers sST2 (OR = 1.960), MMP-2 (OR = 0.805) and NT-proBNP (OR = 0.002) were all independent risk factors for patients with heart failure. Survival analysis results suggested that for patients with HFmrEF, a higher level of plasma sST2 (≥ 0.332 ng/ml at admission) may predict a higher risk of endpoint events and a lower survival rate (P < 0.025).