The relationship between the occurrence of side effects (SEs) and drug factors, such as antiepileptic drugs (AEDs), daily dose, duration of treatment, drug combination pattern, total and free serum concentrations, metabolite per parent level ratio as an index of metabolism ability, and co-medicated drugs except AEDs were evaluated in 227 outpatients with epilepsy. The possible influences of certain physiological and/or the pathophysiological factors were also evaluated. SEs with 19 clinical signs were observed in 66.1% of all patients. There was no definite dose- or serum concentration-dependent increase in the incidence of SEs. Stepwise discriminant function analysis revealed that benzodiazepines (BZN) polytherapy with AEDs produced a higher incidence of somnolence and general fatigue than did any other AED or drug combination. The effects of various drug combination patterns on the incidence of SEs were also evaluated on the basis of observed frequencies. The incidence of somnolence was significantly higher in patients taking phenytoin (PHT) plus carbamazepine (CBZ) therapy, and in patients taking BZN plus either PHT, phenobarbital (PB) or CBZ therapy compared with patients taking either PHT, PB or CBZ therapy. Other responsible drug combination patterns were PHT plus valproic acid (VPA) therapy for mental function impairment, acetazolamide (AZM) polytherapy with PB or PHT for dry mouth, and CBZ plus BZN therapy for constipation. In this study, the stratifying points (occurrence limits) of SEs were detected in various variables such as the number of prescribed drugs, daily dose and serum concentrations. Interestingly, these limits are within the commonly accepted "therapeutic range" or "usual daily dose," and some of these limits shifted down when another AED was co-medicated.(ABSTRACT TRUNCATED AT 250 WORDS)