It is well documented that emotionally arousing experiences are better remembered than mundane events. This is thought to occur through hippocampus-amygdala crosstalk during encoding, consolidation, and retrieval. Here we investigated whether emotional events (context) also cause a memory benefit for simultaneously encoded non-arousing contents and whether this effect persists after a delay via recruitment of a similar hippocampus-amygdala network. Participants studied neutral pictures (content) encoded together with either an arousing or a neutral sound (that served as context) in two study sessions three days apart. Memory was tested in a functional magnetic resonance scanner directly after the second study session. Pictures recognised with high confidence were more often thought to have been associated with an arousing than with a neutral context, irrespective of the veridical source memory. If the retrieved context was arousing, an area in the hippocampus adjacent to the amygdala exhibited heightened activation and this area increased functional connectivity with the parahippocampal gyrus, an area known to process pictures of scenes. These findings suggest that memories can be shaped by the retrieval act. Memory structures may be recruited to a higher degree when an arousing context is retrieved, and this may give rise to confident judgments of recognition for neutral pictures even after a delay.