Effects of the anticonvulsant drugs sodium valproate (NaVP) and ethosuximide (ES) on mouse central (spinal cord and cortical) neurons in primary dissociated cell culture were studied using intracellular recording techniques. Drug effects on two properties of the neurons were assayed: the ability to sustain high frequency repetitive firing (SRF) of sodium-dependent action potentials, a voltage sensitive nonsynaptic membrane property; and the amplitude of responses to iontophoretically applied gamma-aminobutyric acid (GABA), a postsynaptic effect of this inhibitory amino acid neurotransmitter. At concentrations equivalent to the clinically useful therapeutic range in cerebrospinal fluid (6-200 microM), NaVP limited SRF to a few action potentials in both spinal cord and cortical neurons during long (450 msec) depolarizing current pulses. The limitation of SRF was paralleled by use-dependent reduction of maximal rate of rise (Vmax) of the action potentials and prolongation of recovery of Vmax from inactivation. This action was similar to limitation of SRF produced by phenytoin and carbamazepine. The 2-en-metabolite of NaVP, sodium 2-propyl, 2-pentenoate, did not limit SRF at 12 to 120 microM. However, the diphenyl analog of NaVP, sodium diphenylacetate, limited SRF at concentrations between 4.7 to 23.5 microM. ES did not affect SRF at concentrations up to 700 microM. At concentrations of 120 to 1000 microM, including the upper limit of therapeutic range, NaVP did not affect postsynaptic GABA responses in 80% of spinal cord neurons. In the remaining 20%, GABA responses were augmented less than 33%. ES reduced slightly (22%) GABA responses at a high concentration (700 microM). These findings suggest that limitation of SRF may be an important cellular mechanism by which NaVP, but not ES, exerts anticonvulsant efficacy and that neither ES nor NaVP have anticonvulsant action by enhancing postsynaptic GABA action.