We used an item response theory (IRT) measurement model to examine the 28-item Conners Teacher's Rating Scale (CTRS) for the diagnosis of attention-deficit/hyperactivity disorder (ADHD). Using an Israeli sample of 453 pairs of respondents (parents and teachers), we broke down our sample into nonclinical (n = 306) and clinical (n = 147) subsamples. IRT analyses based on a rating scale model revealed structural inadequacies involving the apparent inappropriateness of the 4-point Likert-type scale used by the CTRS. Additionally, IRT analyses were able to shed light on differences in the structure of the latent trait for the ADHD variable as measured by the CTRS for all four groups of respondents. Implications for the diagnosis and treatment of ADHD are presented.