OBJECTIVES: 1) To model long-term clinical and cost-effectiveness of the Diabetes Education and Self-Management for Ongoing and Newly Diagnosed (DESMOND) for people with newly diagnosed Type 2 diabetes, versus usual care in the UK; 2) To consider issues arising in modelling education / behaviour modification programmes and how such modelling can support the development of clinical research. METHODS: The modelling undertakes a long-term cost-utility analysis with evidence from a 12-month multicentre cluster RCT using the Sheffield Type 2 Diabetes Model. Short and long-term outcomes include Hba1c, lipids and systolic blood pressure, patients' weight and smoking status. The model examines long-term use of therapies including oral hypoglycemic agents. Risk / disease progression models based on UKPDS and other evidence are used to estimate incidence of complications, mortality, costs and health-related quality of life. RESULTS: Estimated mean (95% CI) incremental lifetime costs per person (trial based) = ≤ 218 (-≤ 194 to ≤ 758), incremental QALYs= 0.0406 (-0.0283 to 0.1050), and incremental cost/QALY= ≤ 5369. Using current real-world costs, the incremental cost= ≤ 91 (-≤ 321 to ≤ 631) and incremental cost/QALY= ≤2241. Probabilistic sensitivity analysis suggests an 87% likelihood of being cost effective at a ≤20,000/QALY threshold. One-way sensitivity analysis showed results are most affected by effects around weight and smoking, and that DESMOND would likely be cost effective under the conservative assumption of zero effect maintained after year one. CONCLUSIONS: Whilst results suggest that DESMOND is cost effective compared to usual care, further modelling should include: whether maintenance of effect via longer-term top-up education is effective, subgroup analysis of those who respond / do not respond, direct modelling of exercise benefits, and adaptation to education programmes such as Dose Adjustment for Normal Eating (DAFNE) in Type 1 diabetes.