The prevalence of heart failure (HF) in patients with type 2 diabetes (T2DM) is debatable and no data exist concerning the diagnostic value of mid-regional pro-atrial natriuretic peptide (MR-proANP). We aimed to identify HF prevalence and evaluate the diagnostic value of MR-proANP in outpatients followed in two specialized diabetes clinics. HF was pre-defined as HF with reduced ejection fraction (HFrEF) and HF with preserved ejection fraction (HFpEF). The prevalence of HFrEF and HFpEF was 2.4% and 17.5%, respectively. An MR-proANP <60 pmol/L ruled out HFrEF in the total population (n = 806) and in patients reporting dyspnea (n = 311) with a sensitivity of 94.7% and 87.5%, a negative predictive value of 99.7% and 99.0%, a specificity of 39.5% and 33.0%, and a positive predictive value of 3.6% and 3.3%, respectively. In a multivariable model including age, sex, T2DM duration, albuminuria, uncontrolled systolic blood pressure, abnormal electrocardiogram and ischaemic heart disease for diagnosis of HF in patients reporting dyspnea, adding MR-proANP increased the area under the curve from 0.69 to 0.78 (P < 0.001). In conclusion, HFrEF was rare among outpatients with T2DM. MR-proANP rules out HFrEF and contributes independent information relevant to diagnosis of HF in patients reporting dyspnea.