OBJECTIVES: To estimate the economic inefficiency of bad control of Diabetes in terms of direct cost in diabetic patients in a rural area in Spain (Axarquía) during one-year of follow up. METHODS: Cohort study of 228 patients (2 rural health centres sampled in 2005-2006). The perspective analysis was National Health System. Total cost is composed of four items: insulin and oral hypoglycaemic agents, glycosylated haemoglobin, disposable and consumable goods (glucose test strips, needles, and syringes) and primary care visits. The statistical analysis was made by the SPSS package. For quantitative: mean, DS, IC95%; for qualitative: proportions. For inferential statistics: Student's T-test for quantitative variables and test of Chi-square for the qualitative ones. Multivariante analysis considered: age, gender, control grade (ADA, 2006) and interaction gender-control of DM. The ratio cost-effectiveness and incremental cost of control compared with non-control of DM was analized with WinBugs controlled for age and gender. RESULTS: Sixty-seven percent of females; average age: 69 years. The average of prescribed drugs for DM control was 1.4 drugs/per patient (1.03% acarbose; 32.99% sulphonylurea; 64.43% biguanide, 13.40% metiglinide and the 27.32% insuline). The mean of primary care visits was 14.8/per patient/year. The Diabetes control was 68% (HbA1c% <7) with 1.34 drugs per year in controlled patients versus 1.61 drugs in non-controlled (p<0.01). The direct health care cost of diabetic patient was €628/year (CI95%: €576–680). In patients non haemoglobin glycosilated controlled the cost was incremented in €193 per patient and the effectiveness decal in 32%. CONCLUSION: This economic study demonstrates the real possibility to improve the efficiency of our interventions. If we transferred the data found to our area of reference (Málaga) the saving in costs would be superior to €3 million. Health providers and policymakers should use this information in making clinical and policy decisions in order to use resources efficiently