This study explored a new approach to the application of electroencephalogram (EEG) power spectral analysis in the search for an objective, quantitative evaluation of anticonvulsant drug effects on the primate central nervous system. Standardized EEG samples were drawn from the slow wave sleep state. Four rhesus monkeys were adapted to restraining chairs and to prolonged recording in an isolation cubicle. Surgically placed, permanent electrodes provided for the monitoring of sleep-waking states and of bipolar frontal, central and occipital EEG traces. Baseline spectral density data were drawn from standard successive and spaced recording nights, and from post-intramuscular saline injection recording nights. These were compared with identical data drawn 1.5-2.5 h after a single, acute intramuscular administration of 5 compounds, including: diazepam, carbamazepine, valproic acid, phenobarbital and pentobarbital, the latter providing a non-anticonvulsant control. All dose values were low relative to clinical norms, as confirmed by serum samples drawn after EEG samples. Baseline measures showed stability across all non-drug test conditions, particularly those derived from somatosensory cortex. Identical post-drug power spectral changes indicated a unique and significant attenuation of 4-7 c/sec activity bilaterally in somatosensory cortex for all test compounds except pentobarbital, when compared to saline values. Since pentobarbital has only weak anticonvulsant actions, these data suggest a common effect of anticonvulsant drugs on central cortical EEG substrates. Other neurophysiological findings are reviewed which suggest that this effect may stem from an altered cortical response to intrinsic somatosensory thalamo-cortical afferent discharge in sleep, resulting from reduced neuronal excitability.