The aim was to investigate the association between diabetes duration and depressive symptoms in type 2 diabetes. The DIAREG registry used data of a nationwide general medicine practice database (Disease Analyzer, Germany) augmented by prospective data from patient reported outcomes (PRO) including Center of Epidemiological Studies Depression Scale (CES-D: <16 vs. ≥16) and the SF-36. Multiple logistic regression models were used to investigate the association of diabetes duration with PROs. Overall, 1807 type 2 diabetes patients were registered in 108 practices. From 270 (15%) patients complete PRO could be collected, which were similar with respect to age, sex, body mass index, HbA1c, diabetes duration and treatment to patients with incomplete data (p≥0.05). Patients with a longer diabetes duration (reference: <2 years) displayed a significantly increased odds of having no indication of depression (CES-D <16: 66%) (Odds Ratio, 95%CI: 2-<5 years: 5.9, 1.2-29.6; 5-<10 years: 6.2, 1.3-28.7; ≥10 years: 5.6, 1.2-23.1), after adjusting for age, sex, BMI, diabetes complications, antidepressants and glucose-lowering treatment. Patients with <2 years diabetes duration also exhibited a significantly decreased mean mental component score (SF-36: <2 years: 50, 2-<5 years: 69). In type 2 diabetes with a short duration an increased depressive symptom score was observed.