Controversy has arisen about the effectiveness of phenytoin against kindled seizures. It has been suggested that the reports of ineffectiveness could be accounted for by phenytoin being given by an intraperitoneal (i.p.) route in those experiments so that adequate serum concentrations were not achieved. Another possibility for the different results was dissimilar stimulus protocols employed in the various studies. The present study examined these issues. Doses of i.p. phenytoin were studied for their actions against kindled responses elicited with short (1 s) and long (10 s) stimulus trains through hippocampal electrodes. Serial application of the stimuli determined time-action relationships. Dose-dependent effects were demonstrated for all time points examined. There was a consistently greater suppression of kindled motor seizures than limbic behavioral seizures or electrographic seizures. Phenytoin either totally blocked or did not affect the duration of afterdischarges. Actions of phenytoin against responses by short duration stimuli were greater than against long duration stimuli. Additional pharmacokinetic studies compared i.p. versus intravenous (i.v.) phenytoin. After i.p. phenytoin, serum levels peaked later than after i.v. delivery, but were maintained in the 'therapeutic' range longer. The present experiments provide additional support for the idea that kindled seizures are a useful model for complex partial seizures in humans. In addition, they show that major actions of phenytoin are to decrease seizure spread and to elevate afterdischarge thresholds and that the i.p. route is appropriate for assaying the effect of phenytoin against kindled seizures in rats.