Children exposed to valproate monotherapy in utero were evaluated with respect to neurological functioning, behavior, and additional educational needs, and the results were compared with those for age- and gender-matched controls exposed to carbamazepine and children with no prenatal exposure to antiepileptic drugs. We identified from the community-based pregnancy registry of Kuopio University Hospital area (1989-2000) all first-born and school-aged children exposed to valproate (N=13). Neurological and neuropsychological assessments were made clinically, and behavioral problems were assessed with the Conners' Teacher Rating Scale (CTRS). Eight children (62%) exposed to valproate and two (15%) each in the carbamazepine-exposed and nonexposed groups (P=0.022) required educational support. Minor dysmorphic features were noted in eight children (62%) exposed to valproate and in three children (23%) each in the carbamazepine-exposed and nonexposed groups. On CTRS, children exposed to valproate received higher scores, indicating behavioral problems. In our small but population-based study, all children exposed to valproate had minor, and some of them major, cognitive or neurological problems. This difference is clearly observed when assessing each child individually, but the many confounding factors explaining at least part of this difference are difficult to control and avoid in clinical practice. Larger studies with a prospective design are needed to confirm these findings.