OBJECTIVES: Type-2 diabetes in France has increased in prevalence over the past decade and there is a need to examine cardiovascular outcomes based on real-world evidence (RWE). The CANARD model was developed to evaluate the cost-effectiveness (CE) of treatments using data from a RWE French population in the IMS LifeLinkTM Diabetes Cohort database. METHODS: A Markov micro-simulation model was developed in Microsoft Excel 2007® using the United Kingdom Prospective Diabetes Study (UKPDS) risk equations to predict cardiovascular outcomes (MI, stroke, death). Four treatment pathways taken from the IMS LifeLinkTM Diabetes Cohort database were evaluated, which included step-wise treatment switches based on the patient’s HbA1c. Treatments within the pathway included diet and exercise, metformin (MF), oral anti-diabetics (OADs) of sulfonylurea (SU) or dipeptidyl peptidase IV (DDP-IV) in addition to MF, a lipid lowering agent (atorvastatin) in addition to MF and an OAD, and insulin. CE outcomes in 13 subgroups were explored and uncertainty in the model was evaluated with a probabilistic sensitivity analysis (PSA) and expected value of perfect information (EVPI). The model was validated by comparing the predicted results to the database results at 18 months.   RESULTS: The results showed that the treatment pathway including SU with atorvastatin was the most cost-effective in the general population at €2,640/QALY.  The results from the PSA showed that the likelihood of this treatment pathway being cost-effective was 0.61 at willingness-to-pay (WTP) of €20,000/QALY, and 0.56 at a WTP of €50,000/QALY. EVPI at a WTP of €20,000/QALY and €50,000/QALY were €476 per patient and €1,427 per patient respectively. The subgroup results demonstrated that depending on the WTP of the French National Health Service, several treatments pathways may be CE. CONCLUSIONS: Overall, SU with atorvastatin was the most cost-effective treatment for a French type II diabetic population, however, uncertainty was evident and therefore, future research would be of value.