The aim of this observational study was to assess mortality of patients with type 2 diabetes by type of healthcare delivery system, i.e. through specialist centers or generalist doctors, or integrated care. METHODS: The study was conducted at the "Vicentino Ovest" Local Health District in the Veneto Region (north-eastern Italy) from January 1, 2008 to December 31, 2010. Patients with diabetes (≥20 years old) were identified using different public health databases. They were grouped as: patients followed up by specialists at diabetes clinics (DS); patients seen only by their own general practitioner (GP); and patients receiving integrated care (DS-GP). Cox's regression analysis was used to estimate adjusted hazard ratios for available potential predictors of death by level of care. RESULTS: The crude mortality rate was highest in the GP group (26.1 per 1000 person-years), the difference being minimal when compared with the DS group (21.7 per 1000 person-years) and more marked when compared with the DS-GP group (8.8 per 1000 person-years). Patients followed up by their GPs had a 2.7 adjusted RR for mortality by comparison with the DS-GP group. CONCLUSIONS: The findings of the present study could demonstrate that it is safe and cost-effective, after a first specialist assessment at a diabetes service, for low-risk diabetic patients to be managed by family physicians as part of a coordinated care approach, based on the specialist's clinical recommendations; GPs can subsequently refer patients to a specialist whenever warranted by their clinical condition.