Functional MRI (fMRI) studies of memory with coarse resolution of 4 × 4 × 5 mm often fail to demonstrate blood oxygenation level-dependent (BOLD) activation in the hippocampal formation. This failure occurs when nonactivating white matter is averaged with the signal from hippocampal gray matter, attenuating the total BOLD signal from a single voxel due to the "partial volume effect." In this study, we evaluated the suitability of high-resolution fMRI at 3T (voxel size 2 × 2 × 3 mm) for improved visualization of hippocampal activation during memory encoding in 21 healthy and 6 epilepsy subjects. We used a picture encoding task (block design) that involved memorization of indoor and outdoor scenes along with an appropriate resting task. Region-of-interest (ROI) analysis was performed; laterality indices (LIs) were calculated based on hippocampal ROIs (hROIs) or on global medial temporal ROIs (mtROIs). In 19 healthy subjects, robust bilateral BOLD signal changes within both ROIs were noted. The mean LI ± SD for the hROIs is -0.12 ± 0.06 and that for the medial temporal ROIs -0.12 ± 0.05, with correlation between the LIs (r = 0.59, P = 0.009). Good concordance was noted between the surgical outcome and memory lateralization with the fMRI task employed in this study. The preliminary results are encouraging, and with continuing improvements in MRI scanner technology, we expect fMRI of the hippocampal formation at higher resolution to be possible and preferable. Furthermore, these results suggest that a larger study to test the utility of high-resolution fMRI in epilepsy presurgical evaluation is needed. © 2004 Elsevier Inc. All rights reserved.