The phenomenon of temporally graded retrograde amnesia (loss of information acquired before the onset of amnesia) suggests that the hippocampus, and possibly other medial temporal lobe (MTL) structures, have a time-limited role in memory. In three experiments, we made a first attempt to use fMRI to assess activity in the hippocampal region (the CA fields of the hippocampus, the dentate gyrus, and the subiculum) and the adjacent parahippocampal gyrus (parahippocampal, entorhinal, and perirhinal cortices) during recognition memory testing as a function of study-test interval. Experiment 1 (n = 5) demonstrated activity in the hippocampal region and parahippocampal gyrus for targets relative to foils during recognition memory performance following a single study-test delay of about one-half hour. In Experiment 2, 15 participants studied line drawings at each of three different times prior to scanning: one-half hour, 1 day, and 1 week. fMRI data were then collected during recognition memory testing, using targets from all three delays and foils. While an overall effect of targets vs. foils was found in both the hippocampal region and the parahippocampal gyrus, there was no effect of study-test interval on target activity. In Experiment 3 (n = 13), behavioral performance (reaction time and accuracy) was equated across the three delays. Again, no effect of study-test interval was observed. It is possible that the time span sampled in our study (one-half hour to 1 week) was too short to observe changes in activity. Alternatively, activity in the MTL during memory testing may occur even when these structures are not necessary for retrieval.