DSM-5 requires for the diagnosis of attention deficit hyperactivity disorder (ADHD) the presence of at least six symptoms, resulting in three subtypes or presentations. However, the validity of the subtypes remains debated. Our objectives were (a) to investigate whether ADHD is best characterized as a typology, a continuum, or a mixture of both, and (b) to investigate whether ADHD presentations proposed by DSM-5 can be empirically validated. For this purpose, we compared confirmatory factor models, latent profile models and mixture factor models on data from 871 children (5 to 14 years old). The best model was that of two factors (attention deficit and hyperactivity/impulsivity) and four classes. The classes did not represent groups with qualitatively different symptomatic profiles, but a mixture of approximately normal distributions underlying the same severity continuum. These results do not support the categorization of children in subtypes or presentations, but their assessment in a continuum of severity whose upper end could be understood as the clinical manifestation of ADHD.