Objective: Czech guidelines recommend additional therapy with sulfonylurea, insulin or thiazolidinediones (TZD) after metformin monotherapy failure in type 2 diabetes. However TZD in Czech are perceived to represent a cost intensive treatment compared to insulin. The aim of this study was to assess annual direct medical costs/reimbursement in patients treated either with 1) rosiglitazon + metformin, or 2) insulin (monotherapy or combination) from the health care perspective. Further we performed a cost-effectiveness calculation related to HbA1c decrease. Methods: Total 199 patients with completed 12 months TZD or insulin treatment, who failed metformin monotherapy, were included into the analysis. Following data were recorded retrospectively: Medication, consultations related to diabetes, hospitalization (incl. ward type and lengths), devices for insulin application, patient education, selfmonitoring costs, sick-leave and HbA1c values at the beginning and end of the assessed period. Costs were calculated in 2007 prices. Results: Age in both groups was comparable. Insulin patients had significantly higher entry and final HbA1c values (8.3% vs. 7.3% and 6.5% vs. 6.0% p<0.05). Insulin treated had higher mean annual costs compared to TZD (€867 vs. €643 p<0.05); however costs per 1% HbA1c decrease were comparable between groups (€475 vs. €469). Mean total costs for a subgroup of patients who achieved HbA1c of =6.0% (indicating satisfactory compensation) at the end of assessment were also significantly higher in the insulin group (802€ vs. €610 p<0.05) despite similar entry and final HbA1c levels. Costs per 1% HbA1c decrease were comparable also in this subgroup (€375 vs. €366). Conclusion: Insulin or TZD added on after metformin monotherapy failure in type 2 diabetes resulted in comparable costs per 1% HbA1c decrease. Outcomes were robust also for patients who achieved HbA1c levels of =6.0%. Both treatment options result in similar cost-effectiveness from an annual perspective.