OBJECTIVES: To describe and discuss existing pharmacoeconomic models assessing obese or overweight patients being treated with anti-obesity drugs METHODS: Medline and EMBASE were searched for cost-effectiveness or cost-utility-analysis reporting in full decision modelling used. It was eligible for assessment only studies considering Quality Adjusted-Life Years or Life-Year Gained as the final outcome and assessing pharmacotherapy as intervention. Data regarding population, study's perspective, time-horizon and methodology used were described. It was discussed the sensitivity of the model for obesity's co-morbidities such as cardiovacular diseases and type 2 diabetes mellitus and their reliability with clinical practice. In addition, it was explored the risk equations used to populate the model. RESULTS: From 51 studies assessed in full, twelve of them complied with our inclusion criteria being that were identified twelve different models. The most common methodology used was health-state transition modelling followed by life-tables. The vast majority of included studies were carried out in Europe, being 4 of them in United Kingdom. Ten studies considered lifetime horizon and two studies five- and one-year horizon. Diabetics and non-diabetics patients were taken into account in almost all models. Changes in body mass index (BMI) promoted by the pharmacotherapy were correlated to the risk of experiencing coronary heart diseases, stroke or death in ten studies. Furthermore, two models took into account natural BMI increasing due to population ages. Besides BMI, some models were sensitive to changes in total cholesterol, HDL, systolic blood pressure and glycated hemoglobin for diabetics. Finally, equations derived from Framingham Heart Study (FHS) were the most common ones used to populate models and competing risks events were not discussed in most of studies.      CONCLUSIONS: Health-state transition modelling seems to be the most appropriate methodology to assess obesity's pharmacotherapy. Future studies should explore other co-morbidities and competing risks events. Furthermore, equations used to populate models must be coherent to target population characteristics.