OBJECTIVE: To test whether the reported association between pediatric epilepsy and behavioral problems may be distorted by the use of parental proxy report instruments. STUDY DESIGN: Children in the Connecticut Study of Epilepsy were assessed 8-9 years after their epilepsy diagnosis (time-1) with the parent-proxy Child Behavior Check List (CBCL) (ages 6-18 years) or the Young Adult Self-Report (>/=18 years of age). For children <18 years of age, parents also completed the Child Health Questionnaire, which contains scales for impact of child's illness on the parents. The same study subjects completed the Adult Self-Report 6-8 years later (time-2). Sibling controls were also tested. Case-control differences were examined for evidence suggesting more behavioral problems in cases with epilepsy than in controls based on proxy- vs self-report measures. RESULTS: At time-1, parent-proxy CBCL scores were significantly higher (worse) for cases than controls (n = 140 matched pairs). After adjustment for Child Health Questionnaire scales reflecting parent emotional and time impact, only 1 case-control difference on the CBCL remained significant. Self-reported Young Adult Self-Report scores did not differ between cases and controls (n = 42 pairs). At time-2, there were no significant self-reported case-control differences on the Adult Self-Report (n = 105 pairs). CONCLUSIONS: Parent-proxy behavior measures appear to be influenced by the emotional impact of epilepsy on parents. This may contribute to apparent associations between behavioral problems and childhood epilepsy. Self-report measures in older adolescents (>18 years of age) and young adults do not confirm parental perceptions. Evidence suggesting more behavioral problems in children with epilepsy should be interpreted in light of the source of information.