The current study examined how sensory and emotional information is integrated in the mammalian brain. To do so, it used a sensory preconditioning protocol in which rats integrate a tone-light association (sensory information) formed in stage 1 with a light-shock association (emotional information) formed in stage 2 to generate fear responses (freezing) when tested with the tone alone in stage 3. This integration could occur through chaining of the tone-light and light-shock associations during testing with the tone (e.g., tone-->light-->shock); or through formation of a mediated tone-shock association during conditioning of the light. Following previous work, we distinguished between these alternatives by blocking NMDA receptors (using micro-infusions of the antagonist, D-AP5) in the perirhinal cortex (PRh) during stage 2. We found that: 1) after just a few tone-light pairings in stage 1, the D-AP5 infusion disrupted freezing to the tone (but not the light), consistent with integration through mediated learning; 2) after many tone-light pairings in stage 1, the D-AP5 infusion no longer disrupted freezing to the tone, consistent with integration through chaining at test; but 3) in the latter case, the effect of the D-AP5 infusion was restored (i.e., freezing to tone was disrupted) when the interval between stages 1 and 2 increased from 24 hours (standard) to 15 days, suggesting a return to integration via mediated learning. These results show that familiarity and time influence how the brain integrates sensory and emotional information. They are discussed with respect to theories of information processing in the medial temporal lobe.