Background: Dyspnea is a common clinical challenge in patients with type 2 diabetes and may be a sign of heart failure (HF). We sought to evaluate the predictive value dyspnea with and without an echocardiographic substrate in patients with type 2 diabetes without known heart disease. Methods: A total of 724 patients with type 2 diabetes followed at specialized clinics participated in this prospective cohort study. Clinical evaluation, comprehensive echocardiography and follow-up through national registers were performed. An echocardiographic substrate was either left ventricular hypertrophy, increased left atrial size, E/e’ > 15, or LV ejection fraction<50%. The end-points were cardiovascular (CVD)events and all-cause mortality. Results: Median follow-up was 4.8 years [Interquartile range: 4.1, 5.3]for CVD event and 77 patients suffered a CVD event. Dyspnea was significantly associated with CVD event: Hazard ratio (HR): 1.58 (95% confidence interval: 1.01–2.48), p = 0.04. Stratifying by evidence of echocardiographic substrate revealed high risk individuals: CVD event: 0.71 (0.35–1.46), p = NS in patients with dyspnea and no echocardiographic substrate and 2.85 (1.74–4.67), p < 0.001 in patients with dyspnea with echocardiographic substrate). This pattern was similar in multivariable analyses. Also, C-statistics improved from 0.66 (0.60–0.72)to 0.69 (0.63–0.75), p < 0.001 and net reclassification index was 27.5%(5.0–50.0), p = 0.01 for CVD event. The results were similar for all-cause mortality except dyspnea was only a borderline significant predictor. Conclusion: In patients with type 2 diabetes complaining of dyspnea, identifying an echocardiographic substrate - thus indicating patients with HF - accurately stratifies patients with increased risk of CV events and all-cause mortality.