Rabbits received classical conditioning of the nictitating membrane response (NMR) in a trace conditioning paradigm. In this paradigm, a 250-ms tone conditioned stimulus (CS) occurs, after which there is a 500-ms period of time in which no stimuli occur (the trace interval), followed by a 100-ms air puff unconditioned stimulus (UCS). In Experiment 1, lesions of the hippocampus or cingulate/retrosplenial cortex disrupted acquisition of the long-latency or adaptive conditioned response relative to unoperated controls and animals that received neocortical lesions that spared the cingulate/retrosplenial areas. When animals with hippocampal or cingulate/retrosplenial lesions were switched to a standard delay paradigm in which the CS and UCS were contiguous in time, they acquired in about the same number of trials as naive rabbits. In a second experiment multiple-unit activity in area CA1 of the hippocampus was examined during acquisition of the trace conditioned response (CR). Three groups of animals were tested: animals that had a 500-ms trace interval (Group T-500), animals that received explicitly unpaired presentations of the CS and UCS (Group UP), and animals that underwent conditioning with a 2,000-ms trace interval (Group T-2000). Animals in Group T-500 acquired the CR in about 500 trials. Early in training, and well before any CRs occurred, there was a substantial increase in neuronal activity in the hippocampus that began during the CS and persisted through the trace interval. There was also an increase in the UCS period that modeled the amplitude-time course of the behavioral unconditioned response. Later in conditioning as CRs emerged, there was no longer neuronal bursting throughout the CS + trace period. Rather, the activity shifted to later in the trace interval and formed a model of the amplitude-time course of the behavioral CR. Activity during the UCS period was similar to that seen earlier in conditioning. Animals in Group UP showed no behavioral conditioning and no increase in neuronal activity. Animals in Group T-2000 showed no long-latency behavioral conditioning and no increase in neuronal activity. The data are discussed in terms of the role of the hippocampus in conditioning during situations in which the CS and UCS are not contiguous in time.