1. Focal cortical epilepsy was investigated by applying tetanic stimulation repeatedly (100 Hz. 2 s in duration, once every 10 min, 10 episodes) to layer III association fibers in rat piriform cortex slices and recording both extracellular and intracellular responses from the endopiriform nucleus. To promote excitability, piriform slices were incubated in artificial cerebrospinal fluid (ACSF) containing 0.9 mM Mg2+ and 5 mM K+, at an initial temperature of 10–12 degrees C, which was allowed to warm passively to room temperature. 2. Responses recorded extracellularly in the endopiriform nucleus consisted of two types: weak stimulation evoked an early-occurring, small-amplitude, negatively deflecting potential; strong stimulation evoked a more complex response comprising both an early potential of maximal amplitude and a later-occurring epileptiform potential of greater amplitude and longer duration. Late-occurring epileptiform potentials were not observed in slices incubated in ACSF at room temperature. 3. Both the early potential and the late-occurring epileptiform responses were abolished by the non-N-methyl-D-aspartic acid (non-NMDA) subtype of glutamate receptor blocker, 6-cyano-7-nitroquinoxaline-2,3-dione (CNQX; 10 microM). Application of D(-)-2-amino-5-phosphonopentanoic acid (APV; 50 microM) to block NMDA receptors was without effect on the early potential but diminished the late-occurring epileptiform potential. The late-occurring potential was unable to follow stimulation delivered at a frequency of 1 Hz. These results suggest that the early potential was generated monosynaptically and dependent solely on the activation of non-NMDA receptors, whereas the late-occurring epileptiform potential was polysynaptic in origin and possessed both a CNQX- and an APV-sensitive component. 4. Responses increased progressively in both amplitude and duration after tetanic stimulation. The threshold intensity required to evoke the complex dual-component potential was reduced by tetanic stimulation. An increase in multiunit spiking activity, indicating an increase in synchronous discharges, was also observed. A residual potential could be evoked in the presence of CNQX (10 microM) after the tetanic stimulation procedure. 5. Spontaneous discharges occurred as early as after the first episode of tetanic stimulation and persisted for the duration of the experiment. Spontaneous discharges were abolished by either CNQX or by a fourfold increase in extracellular Mg2+ concentration, the latter reversibly. APV reduced the frequency of spontaneous discharges by 38.6 +/- 9.3% (mean +/- SE). The conventional anticonvulsant drug 5,5-diphenylhydantoin, the benzodiazepine receptor agonist midazolam, and the benzodiazepine receptor antagonist flumazenil were without effect on the frequency of spontaneous discharges. Evoked responses were also unaffected by either 5,5-diphenylhydantoin or midazolam. Slices not exposed to cold ACSF, although demonstrating potentiation of evoked responses after tetanization did not produce spontaneous epileptiform discharges. 6. Intracellular recordings from endopiriform neurons revealed the cellular correlates of the extracellular responses. Weak stimulation evoked a small-amplitude depolarizing potential. Increasing the intensity of stimulation increased the amplitude of this response and also evoked a second depolarizing potential of greater amplitude occurring at variable latencies. Maximal stimulation evoked an action potential. After tetanic stimuli, responses resembling a paroxysmal depolarizing shift consisting of a depolarizing potential with superimposed multiple action potentials were evoked reliably. Passive membrane properties after repeated tetanic stimulation were not different when compared with control. 7. This novel model of in vitro focal cortical epilepsy has many features characteristic of conventional kindling including 1) progressive nature; 2) reduced threshold to evoke discharges; and 3) persist