Although the child behavior checklist (CBCL) and the teacher's report form (TRF) were not designed for diagnosing psychopathology in children with chronic illnesses, they have become extensively used research tools to assess behavioural problems in paediatric populations, including children with epilepsy. When applied to children with epilepsy, items like "staring blankly" or "twitching" can be rated on the basis of seizure features rather than behaviour and, hence, render behavioural scores ambiguous. The aims were detection, and evaluation of the impact, of CBCL and TRF items eliciting ambiguity when applied to children with "epilepsy only" (idiopathic or cryptogenic epilepsy, attending normal schools). Experts identified items that give rise to interpretational ambiguity of the ratings in epilepsy. By treating ratings on these items as missing values, their effect was evaluated in CBCL and TRF scores of 59 schoolchildren with "epilepsy only" and age and gender matched healthy classmates. Seven items of the CBCL gave rise to ambiguity of which items 5 co-occur on the TRF. Rescoring reduced psychopathology scores in children with "epilepsy only", but not in those of healthy children: the percentage of patients trespassing the clinical cut off score, on at least one of the subscales, reduced from 46 to 23% on the CBCL and from 18 to 15% on the TRF. Parents and teachers run the risk of confusing behaviour and seizure features when filling out the CBCL and TRF. In "epilepsy only", prevalence estimates of psychopathology based on the CBCL and TRF, should be considered with some reserve.