OBJECTIVES: To estimate and compare costs and effectiveness of different educational interventions at PRODIACOR study. METHODS: PRODIACOR is a prospective study (three years), randomized controlled trial, which aims to improve the quality of care for people with type 2 diabetes, optimizing the use of resources and prevent complications, thus reducing morbidity and socioeconomic cost. Includes 4 groups (control, educated patients, doctors educated and educated patients and physicians) of 9 physicians and 117 patients each. Clinical and metabolic changes were recorded in an ad-hoc form (annually and semiannually). Costs and utilization rates for drugs, consultations and practices were obtained from the administrative dataset of the coverage institutions involved. The intervention was approved by independent ethics committee. We performed a descriptive and inferential statistical analysis, verifying differences in means and proportions using t test, ANOVA and Chi square. RESULTS: Clinical and metabolic: although there was improvement of several registered indicators no significant differences in the percentage of patients at goal between groups were found. Cost: in the period before the intervention the total average expenditure for each patient was AR$ 1,848 (66% drugs, 26% practices and 8% consultations). No significant differences between all the intervention groups were found. After the first year all groups significantly increased total average expenditure (Group 1: 259%; Group 2: 39%, Group 3: 59%, Group 4: 61%). The costs then progressively decreased in all groups and at the third year the total average expenditure was AR $1551 (72% drug, 26% practices and 2% consultations). Group 3 had the lowest average total expenditure (AR$ 925), followed by Group 4 (AR$ 1264), Group 1 (AR$ 1881) and Group 2 (AR$ 2133). CONCLUSIONS: Although preliminary, these results suggest that educational interventions improve clinical indicators and reduce direct medical costs associated with diabetes with a satisfactory cost-effectiveness.