OBJECTIVES: To compare clinical outcomes and the health care costs across two cohorts of uncontrolled diabetic patients who initiated treatment with Sulfonamides or Thiazolidinediones (SU/TZD) or Dipeptidyl Peptidase 4 (DPP-4) Inhibitors in a clinical practice setting. METHODS: A retrospective analysis using a large administrative database and a clinical registry containing laboratory results of three Italian Local Health Units was performed. The index-period ranged from July, 2008 and June, 2010. Patients were treatment naïve to SU/TZD or to DPP-4, but already treated with other oral antidiabetic agents. Demographics, concomitant therapies, Charlson comorbidity index, glycemic and lipid control level and previous hospitalizations were assessed at baseline. Adherence was measured by Medication Possession Ratio (MPR). We calculated unadjusted rates and used a Poisson regression model to estimate risk ratios for diabetes-related hospitalizations occurred during the 18-months follow-up period.  Total annual costs included all the pharmacological treatments and the direct costs due to hospitalizations and outpatient services. RESULTS: We identified 1384 patients treated with SU/TZD and 199 treated with DPP-4. DPP-4 patients were significantly younger (mean age 59.2 years and 65.0 years; p<0.001) and with less previous hospital discharges for diabetes-related diseases. Baseline mean HbA1c was  8.1% for SU/TZD and 8.2% for DPP-4 patients. DPP-4 naïve resulted more adherent  (MPR≥80%) than SU/TZD  naïve (70.9% and 55.8%; p<0.001). The SU/TZD group showed a significant increased risk of diabetes-related hospitalizations (unadjusted rate was 9,17 vs 3,47 per 100 person-years, p=0.002; adjusted incidence rate ratio 1.83; p=0.028). The higher hospitalization rate resulted in higher total annual direct costs per patient (€2.719 vs €2.462 of those treated with DPP-4). CONCLUSIONS: Results indicate that uncontrolled diabetic patients who initiated treatment with DPP-4, compared with those initiating with SU/TZD, are associated with a reduced risk of diabetes-related hospitalizations and consequently with a lower overall total annual cost per patient.