Thiazolidinediones (TZDs) can cause weight gain and fluid retention in some patients. In most cases, fluid retention is expressed as mild hemodilution. The incidence of clinically evident edema is relatively uncommon. In large clinical trials with rosiglitazone and pioglitazone, the frequency of edema in TZD-treated patients was about 3 to 4 times higher than in placebo-treated patients. The precise mechanisms responsible for weight gain, fluid retention, and edema associated with TZD therapy are unclear but appear to be both dose- and host-related. Weight gain is most likely multifactorial, reflecting increased body fat and fluid retention. Available data are conflicting and do not completely support the concept that increased body weight and decreased hemoglobin/hematocrit are linked with evidence of fluid retention and hemodilution. As uncommon as edema is, new-onset heart failure is even less common in patients treated with a TZD. In controlled clinical trials, the frequency of congestive heart failure (CHF) was identical in rosiglitazone- and placebo-treated patients. The incidence of CHF is higher in patients receiving combination therapy with insulin and a TZD. Patients in the insulin-treated population who develop CHF tend to be older, have a longer history of type 2 diabetes mellitus, and have risk factors for heart failure in addition to diabetes. TZDs do not necessarily require discontinuation in patients who develop fluid retention or weight gain. Mild fluid retention can be treated by decreasing the TZD dose and/or adding a diuretic. Patients who are taking a TZD should be monitored for signs and symptoms of CHF, including excessive weight gain, edema, and dyspnea. Patients with New York Heart Association (NYHA) class I or II CHF can be treated with TZDs. Therapy should be initiated at low doses and slowly titrated to the lowest effective dose. If CHF worsens or becomes refractory to treatment, it may be necessary to discontinue the TZD. Diagnoses of NYHA class III and IV CHF were not studied in clinical trials of TZDs, and thus TZDs are not recommended for patients with CHF of this severity.