BackgroundRating scales are often used to identify children with potential Attention-Deficit/Hyperactivity Disorder (ADHD), yet there are frequently discrepancies between informants which may be moderated by child characteristics. The current study asked whether correspondence between parent and teacher ratings on the Strengths and Weakness of ADHD symptoms and Normal behaviour scale (SWAN) varied systematically with child language ability.MethodParent and teacher SWAN questionnaires were returned for 200 children (aged 61–81 months); 106 had low language ability (LL) and 94 had typically developing language (TL). After exploring informant correspondence (using Pearson correlation) and the discrepancy between raters, we report inter-class correlation coefficients, to assess inter-rater reliability, and Cohen’s kappa, to assess agreement regarding possible ADHD caseness.ResultsCorrelations between informant ratings on the SWAN were moderate. Children with LL were rated as having increased inattention and hyperactivity relative to children with TL; teachers, however, rated children with LL as having more inattention than parents. Inter-rater reliability of the SWAN was good and there were no systematic differences between the LL and TL groups. Case agreement between parent and teachers was fair; this varied by language group with poorer case agreement for children with LL.ConclusionChildren’s language abilities affect the discrepancy between informant ratings of ADHD symptomatology and the agreement between parents and teachers regarding potential ADHD caseness. The assessment of children’s core language ability would be a beneficial addition to the ADHD diagnostic process.Electronic supplementary materialThe online version of this article (doi:10.1186/s12888-017-1300-8) contains supplementary material, which is available to authorized users.