Recent studies have shown that the integrity of the entorhinal cortex (EC) is not required for simple contextual conditioning. In background contextual conditioning, i.e., when a phasic cue is present during training, the involvement of the EC is still a matter of debate. Therefore, the present work further examines whether the EC is required for background contextual conditioning using a tone as the phasic cue. Rats sustaining either excitotoxic lesions of the EC or sham-lesions were trained with one of two procedures differing with respect to the predictive value of the tone: a paired procedure in which the tone perfectly predicts shock occurrence and overshadows context, and an unpaired procedure in which the predictive value of the tone is reduced. Conditioned fear was assessed by freezing responses during conditioning, reexposure to the training context, and reexposure to the tone in a new context. Postshock freezing was reduced in rats with entorhinal lesions. In all rats trained with the paired procedure, freezing to the context was low and freezing to the tone was high, suggesting that the tone has overshadowed the context during the conditioning session. The reverse pattern was observed with the unpaired procedure in sham-operated rats. In rats with entorhinal lesions trained with the unpaired procedure, freezing responses to the context was markedly reduced. In a new context, however, entorhinal-lesioned rats showed higher freezing scores than those of sham-lesioned rats. Freezing to the tone was unaffected by the lesion irrespective of the tone's predictive value. As a whole, these results support the notion that the EC is required for normal background contextual freezing.