The dopamine transfer deficit model of attention deficit hyperactivity disorder (ADHD) is compared and contrasted with the existing dynamic developmental theory and the extended temporal difference (TD) model. The first two both identify learning deficits as a key problem in ADHD, but this mechanism would seem at least as likely to cause other neurodevelopmental disorders. Learning deficits also do not provide a natural account of ADHD performance worsening in longer trials, nor of stimulant action, nor of the high rate of test–retest inconsistency. The extended TD model is based on the quite different premise that there are very many underlying causes for the disorder, and computational simulation of it has supported the existence of two subtypes that involve no dopamine transfer deficit, as well as several subtypes that do. The three models have markedly different ways of accounting for key aspects of ADHD, including intra-individual variability, stimulant effects, and extinction deficits. The most important shortcoming of all the models is their limited treatment of state-regulation processes, which are very important in ADHD; this is partially addressed by the extended TD model’s proposed dopamine appetite.