Several studies have examined the predictions of two models (i.e., the polygenic multiple threshold (PMT) model and the constitutional variability (CV) model) developed to explain sex differences in the prevalence of attention-deficit/hyperactivity disorder (ADHD), and the results of these studies are conflicting. Although there is substantial evidence that two distinct, moderately correlated symptom dimensions (inattention and hyperactivity–impulsivity) underlie ADHD symptoms, all of these studies have examined ADHD as a unidimensional construct. Therefore, we tested the PMT and CV models for the two ADHD symptom dimensions separately. Dizygotic twins or siblings of girls with ADHD had a higher number of ADHD symptoms than twins or siblings of boys with ADHD. This evidence for the PMT model and against the CV model was found for both inattention and hyperactivity–impulsivity, but became weaker for hyperactivity–impulsivity as symptom severity increased. © 2003 Wiley-Liss, Inc.