Background The aims were to determine whether anxious depression, defined by latent class analysis (LCA), predicts cardiovascular outcomes in type 2 diabetes and to compare the predictive power of anxious depression with Diagnostic & Statistical Manual Versions IV and 5 (DSM‐IV/5) categories of depression and generalized anxiety disorder (GAD). Methods Prospective observational study of 1,337 type 2 participants. Baseline assessment with the 9‐item Patient Health Questionnaire and the GAD Scale; LCA‐defined groups with minor or major anxious depression based on anxiety and depression symptoms. Cox modeling used to compare the independent impact of: (1) LCA anxious depression, (2) DSM‐IV/5 depression, (3) GAD on incident cardiovascular events and deaths after 4 years. Results LCA minor and major anxious depression was present in 21.9 and 7.8% of participants, respectively, DSM‐IV/5 minor and major depression in 6.2 and 6.1%, respectively, and GAD in 4.8%. There were 110 deaths, 31 cardiovascular deaths, and 199 participants had incident cardiovascular events. In adjusted models, minor anxious depression (Hazard ratio (95% confidence intervals): 1.70 (1.15–2.50)) and major anxious depression (1.90 (1.11–3.25)) predicted incident cardiovascular events and major anxious depression also predicted cardiovascular mortality (4.32 (1.35–13.86)). By comparison, incident cardiovascular events were predicted by DSM‐IV/5 major depression (2.10 (1.22–3.62)) only and cardiovascular mortality was predicted by both DSM‐IV/5 major depression (3.56 (1.03–12.35)) and GAD (5.92 (1.84–19.08)). Conclusions LCA‐defined anxious depression is more common than DSM‐IV/5 categories and is a strong predictor of cardiovascular outcomes in type 2 diabetes. These data suggest that this diagnostic scheme has predictive validity and clinical relevance.