Previous studies of phenytoin neuropathy in selected groups of chronic epileptic patients on polytherapy have indicated a widely varying incidence of clinical or electrophysiological abnormalities. In 51 previously untreated epileptic patients followed prospectively on phenytoin or carbamazepine monotherapy, assisted by blood level monitoring, for 1-5 years we found no clinical evidence of neuropathy. Eighteen per cent of the phenytoin group and none of the carbamazepine group had mild electrophysiological changes (abnormalities of sensory action potentials or sensory conduction). In the former group the occurrence of the electrophysiological abnormalities was possibly related to previous exposure to high phenytoin or low folate levels or both. In 10 chronic epileptic patients we demonstrated reversible slowing of sensory nerve conduction during phenytoin intoxication. In six selected epileptic patients on chronic barbiturate monotherapy we found clinical evidence of neuropathy in two and electrophysiological abnormalities in five, including reversible slowing of sensory conduction during intoxication in one. This suggests that barbiturate drugs may, like phenytoin, also contribute to anticonvulsant neuropathy. Careful monitoring of single drug therapy with avoidance of acute toxicity may reduce the risk of chronic anticonvulsant neuropathy.