The effects of epileptiform bursts on hippocampal excitability were examined in the CA3 region of guinea-pig hippocampal slices. Partial blockade of gamma-aminobutyric acidA (GABAA)-mediated inhibition by 500 IU/ml penicillin produced low frequency (2-4 Hz) "pro-convulsant" field potential oscillations. Normal spontaneous activity recovered less than 30 min after the penicillin was rinsed out providing bursting was prevented. Synchronized bursting rarely began on its own even after 1 h in penicillin 500 IU/ml, but could be initiated in most slices after one to eight all-or-none bursts were evoked by low-intensity, low-frequency (0.2-0.25 Hz) stimuli. Spontaneous bursting, once initiated, persisted for at least 1 h without further stimulation suggesting that a small number of bursts produced a long-lasting increase in excitability. Bursts disappeared more slowly than anticipated after convulsants were rinsed out and were followed by "post-burst" oscillations with different frequency characteristics than proconvulsant oscillations which persisted for at least 4 h. Selective augmentation of evoked N-methyl-D-aspartate excitatory postsynaptic potentials appeared to be the critical first step in the initiation of bursting. The specific N-methyl-D-aspartate antagonist, 2-amino-5-phosphonovaleric acid (50-100 microM), only partially suppressed pro-convulsant oscillations in partially disinhibited slices but completely prevented stimulus-triggered spontaneous bursting and prolonged hyperexcitability. Although N-methyl-D-aspartate receptors were necessary for the induction of bursting in partially disinhibited slices, they were not required to initiate bursting after more complete disinhibition. However, when 2-amino-5-phosphonovaleric acid was applied prior to and during perfusion with 2000 IU/ml penicillin, spontaneous bursts occurred at long, irregular intervals and lacked afterdischarges. These bursts rapidly disappeared upon penicillin washout and were not followed by persistent post-burst oscillations. N-methyl-D-aspartate antagonists applied only after bursts already established in penicillin blocked the afterdischarges but did not reduce the burst frequency. These observations indicate that epileptiform bursts can produce long-lasting, hippocampal hyperexcitability. The induction of these plastic changes requires N-methyl-D-aspartate receptor activation which then enhances both N-methyl-D-aspartate and non-N-methyl-D-aspartate receptor mechanisms. Furthermore, N-methyl-D-aspartate excitatory postsynaptic potentials can participate in triggering spontaneous bursts but this role is masked once plasticity has occurred. Partial disinhibition produces a pro-convulsant state which does not induce long-lasting changes in hippocampal excitability but renders the neuronal network vulnerable to develop persistent epileptiform bursting with small additional excitatory inputs.(ABSTRACT TRUNCATED AT 400 WORDS)