Most patients who receive anticonvulsants after traumatic brain injury are treated with the sedative anticonvulsants phenytoin and/or phenobarbital, or perhaps primidone. However, there is considerable evidence demonstrating that these medications have a deleterious effect on cognitive function. Thus, in a rehabilitation setting, alternatives should be sought. Carbamazepine has been found to be relatively free of such effects, and would be an optimum alternative if seizure control were comparable. We have studied the effects of withdrawing phenytoin, phenobarbital and primidone, and using carbamazepine as the primary anticonvulsant in 27 patients at the Greenery Rehabilitation and Skilled Nursing Center for whom ongoing anticonvulsant treatment was considered to be necessary due to previous seizures or a high risk of the occurrence of seizure. We compared a 3 month baseline period (just prior to carbamazepine introduction or sedative anticonvulsant tapering), to a 3 month post-withdrawal period immediately following sedative anticonvulsant withdrawal, when carbamazepine was the sole anticonvulsant. In 20 out of 21 patients in whom carbamazepine replaced sedative anticonvulsants seizure control was essentially similar or somewhat improved. In only one patient did the substitution with carbamazepine result in a loss of seizure control. Six patients were initially receiving carbamazepine in combination with phenytoin and/or phenobarbital. The removal of phenytoin and phenobarbital, leaving carbamazepine as sole therapy, resulted in improved seizure control in three patients and no change in the other three. In the light of carbamazepine's reportedly less detrimental effects on cognitive function and behaviour in other patient populations, it should perhaps be considered as a first line anticonvulsant, especially for patients in rehabilitation settings.