The aim of this study was to determine the discriminative value of irisin for acutely decompensated heart failure (ADHF) in type 2 diabetes mellitus (T2DM) patients with chronic HF. We included 480 T2DM patients with any phenotype of HF and followed them for 52 weeks. Hemodynamic performances and the serum levels of biomarkers were detected at the study entry. The primary clinical end-point was ADHF that led to urgent hospitalization. We found that the serum levels of N-terminal natriuretic pro-peptide (NT-proBNP) were higher (1719 [980–2457] pmol/mL vs. 1057 [570–2607] pmol/mL, respectively) and the levels of irisin were lower (4.96 [3.14–6.85] ng/mL vs. 7.95 [5.73–9.16] ng/mL) in ADHF patients than in those without ADHF. The ROC curve analysis showed that the estimated cut-off point for serum irisin levels (ADHF versus non-ADHF) was 7.85 ng/mL (area under curve [AUC] = 0.869 (95% CI = 0.800–0.937), sensitivity = 82.7%, specificity = 73.5%; p = 0.0001). The multivariate logistic regression yielded that the serum levels of irisin < 7.85 ng/mL (OR = 1.20; p = 0.001) and NT-proBNP > 1215 pmol/mL (OR = 1.18; p = 0.001) retained the predictors for ADHF. Kaplan–Meier plots showed a significant difference of clinical end-point accumulations in patients with HF depending on irisin levels (<7.85 ng/mL versus ≥7.85 ng/mL). In conclusion, we established that decreased levels of irisin were associated with ADHF presentation in chronic HF patients with T2DM independently from NT-proBNP.