Attention deficit/hyperactivity disorder (ADHD) involves clinically heterogeneous problems including attention deficits, behavioural hyperactivity and impulsivity. Several animal models of ADHD have been proposed, ranging from models with neurotoxic lesions to genetically manipulated animals. An ADHD model is supposed to show phenomenological similarities with the disorder, i.e. it should mimic the three core symptoms (face validity). A model should also conform to an established or hypothesized pathophysiological basis of the disorder (construct validity). Finally, an animal model should be able to predict previously unknown aspects of the neurobiology of ADHD or to provide potential new treatments (predictive validity). The currently proposed models are heterogeneous with regard to their pathophysiological alterations and their ability to mimic behavioural symptoms and to predict response to medication. This might reflect the heterogeneous nature of ADHD. Since the knowledge about the biology of ADHD from human studies is limited, one cannot at present decide which model best represents ADHD or certain ADHD subtypes. Animal models with good face and predictive validity may be useful for investigations of the underlying biological substrates of ADHD. At present, the models in use should be described as animal models of ADHD-like symptoms rather than models of ADHD.