To evaluate the use of cerebrospinal fluid (CSF) signal nulling in MR T2 measurements of the hippocampus in normal control subjects and patients with temporal lobe epilepsy (TLE), dual-echo acquisitions covering the whole brain were used. T2 relaxation times were estimated in 12 standard Eurospin II MR test objects and in the hippocampi of 10 control subjects, using T2 maps constructed from conventional spin-echo (CSE), fast spin-echo (FSE), and fast FLAIR (FF) dual-echo sequences on a 1.5-T MR scanner. Hippocampal T2 values (HCT2) were measured on contiguous coronal 5-mm slices throughout the antero-posterior extent of each hippocampus in the 10 controls and 12 TLE patients, using both CSE and FF. Scan-rescan reproducibility in Eurospin II standard MR test objects was high for all sequences. There was a good correlation between T2 values from CSE, FF, and FSE sequences in test objects and in control hippocampi. In controls, the coefficient of variation of mean HCT2 values differed between slice positions, but was lowest for FF, followed by CSE data. The intrarater coefficient of reliability between repeated measurements in control subjects was lowest for FF HCT2, at 2.3%. The interrater coefficient of reliability for CSE HCT2 measurements in controls (4.8%) was slightly lower than the interrater coefficient for FF HCT2 (5.4%). HCT2 measurement with both CSE and FF identified abnormal values in the same 10 hippocampi of 12 patients. Hippocampal dual-echo T2 relaxometry using CSF nulling is reliable in control subjects, and identifies the abnormal hippocampi in patients with TLE. The increases in hippocampal T2 signal demonstrated using FF HCT2 measurements are unlikely to be partial volume effects from CSF.