Evidence suggests that extinction, the suppression of a learned response to a Pavlovian signal that is produced by exposure to the signal alone after conditioning, is a consequence of new inhibitory learning. However, it has been proposed that extinction given immediately after conditioning reflects memory 'erasure'. Using contextual fear conditioning, we examine the nature of extinction further using a novel behavioral paradigm that probes for the absence or presence of a memory. Rats received a context paired with one of three different shock intensities (0.8, 1.2 or 1.6 mA) and then received extinction either immediately (15 min) or after a delay (24 h). Spontaneous recovery was roughly equivalent in the immediate and delayed extinction groups when they were tested at 24 h after extinction. To further test the status of the original memory trace, we exploited the effect that only the first, but not second, learning of contextual fear requires N-methyl-D-aspartate receptors (NMDArs) in the dorsal hippocampus [M.J. Sanders & M.S. Fanselow (2003) Neurobiology of Learning and Memory 80,123-129]. Here we use this property of second learning to determine if memory of an immediately extinguished fear also persists. Rats received bilateral infusions of the NMDAr antagonist DL-2-amino-5-phosphonopentanoic acid into the dorsal hippocampus prior to training in a novel second context. Memory for the second learning is not affected by NMDAr blockade in either group, suggesting that the extinguished memory is not erased but inhibited. Overall, the results provide little evidence that extinction conducted immediately after conditioning destroys or erases the original memory trace.