In a survey of the red cell folate status of 200 patients with epilepsy, compared to 72 controls, we found that median red cell folate levels were reduced significantly in patients treated with phenytoin (p<0.01) or carbamazepine (p<0.001) alone. Patients taking more than one drug had reduced levels also (p<0.001), but in patients treated with sodium valproate alone there was no significant decrease in red cell folate levels compared to controls. Twenty-two per cent of patients in the group taking more than one drug had reduced levels of red cell folate compared with 17 per cent of those taking carbamazepine alone, 13 per cent of those taking phenytoin only, and 9 per cent of those taking sodium valproate only. Dietary folate intake was significantly reduced in all the patient groups compared with controls (p<0.001 for the carbamazepine and phenytoin groups, p<0.01 for the polypharmacy and sodium valproate groups); a significant correlation, between red cell folate levels and dietary folate was not established. Significant negative relationships were established between carbamazepine dose (r=0.35, p<0.01) or serum level (r=-0.27, p<0.05) and red cell folate level in patients on one drug only. The correlation between dose or serum level-of phenytoin and red cell folate level was also negative but did not reach significance. Our findings show that all anticonvulsant drugs interfere with folate metabolism. While the effect is greatest with drugs which induce microsomal liver enzymes, low levels of folate also occurred in patients taking the non-enzyme inducer sodium valproate. Although a significant relationship between diet and red cell folate was not established, dietary folate could be a further contributory factor. © 1987, Oxford University Press.