Background: Thiazolidinediones (TZDs) are widely used antidiabetic agents, but there is great concern and conflicting reports over their possible effect on cardiovascular morbidity, especially in patients with heart failure (HF). Methods: Using 2000-2005 Taiwan's National Health Insurance (NHI) claims data, this population-based, retrospective cohort study investigated if there was an association between the cumulative TZD dose and clinical outcomes in type 2 diabetic patients recently hospitalized for HF. Study outcomes were death, first all-cause readmission, and first readmission for HF. Cox proportional hazard models were used to analyze the association between TZD versus sulfonylurea (SU) use and these outcomes. Results: Out of a total of 8139 eligible patients, 224 were taking TZD (65.63% female; mean [SD] age, 68.30[10.60] years) and 7915 were taking SU (55.10% female; 70.30[10.50] years). Patients taking TZD were at higher risk for readmission for HF (59 cases; HR 1.58 (95% confidence interval, 95%CI 1.44-1.72)), all-cause readmission (147 cases; 1.40 (1.30-1.70)), and death (103 cases; 2.23 (1.58-3.14)). The higher the cumulative exposure to TZD, the greater the risk of HF readmission, all-cause readmission, and death. Conclusion: Among diabetic patients who had been hospitalized for HF, TZD users were at significantly greater risk for readmission for HF, all-cause readmission, and death than SU users. The risk of all adverse clinical outcomes appeared to highly relate to cumulative exposure to TZD. These findings provide empirical evidence supporting the latest black box warnings issued by the United States Food and Drug Administration in August, 2007 advising that TZD not be prescribed for diabetic patients with symptomatic heart failure.