The objective of this study was to examine the effect of carbamazepine and valproate monotherapy on bone mineral density in children. Femoral neck area bone mineral density was measured by dual-energy x-ray absorptiometry in 31 healthy children and 33 children with idiopathic epilepsy treated with either carbamazepine (n = 17) or valproate (n = 16) for more than 6 months. There were no significant differences between the control and study patients in age, height, weight, and physical activity. No patient had dietary restrictions or neurologic impairment. Serum levels (as mean +/- S.D.) of valproate and carbamazepine were 53.75 +/- 23.94 microg/mL and 6.26 +/- 2.00 microg/mL, respectively, and the duration of treatment for each drug was 24.38 +/- 10.58 months and 31.76 +/- 16.33 months, respectively. Calcium intake in the diet was similar in both the control and study groups. In the valproate-treated group, 25% of the patients were hypocalcemic, 6% had elevated alkaline phosphatase levels, and 50% were hypophosphatemic. In the carbamazepine-treated group, 17.6% of the patients were hypocalcemic and 35.3% were hypophosphatemic. Children treated with valproate had 31.9% reduction in bone mineral density at the femoral neck area (P < 0.05); the 20% reduction in bone mineral density in this anatomic location in carbamazepine-treated children was not significant. In conclusion, valproate monotherapy, but not carbamazepine therapy, significantly reduces femoral neck area bone mineral density in children with idiopathic epilepsy.