Obsessive-compulsive disorder (OCD) is a severely disabling psychiatric disorder that is characterised by the presence of obsessions and / or compulsions and affects approximately 2.5% of the general population. The underlying neurobiology is at present not well understood and treatment options for OCD are often of limited efficacy. There is therefore a real need to obtain more knowledge regarding the pathophysiology of OCD and to develop new anti-OCD medication. One way of achieving these aims is to generate animal models for OCD. Naturally occurring animal models of OCD such as acral-lick dermatitis in dogs represent possibly the best models in terms of symptomology, predictive validity and similarity of inducing conditions. Unfortunately, these advantages are offset by the associated high cost of obtaining a sample population. The alternative is to turn to laboratory-based models and these can be classified into behavioural (e.g. food restriction-induced wheel running leading to body weight loss), pharmacological (e.g. 5-HT agonist-induced loss of spontaneous alternation; chronic quinpirole-induced compulsive checking) and genetic (e.g. HoxB8 knock-out mice) subtypes. Each of the different models presents unique advantages and disadvantages. Of the 4 lab-based models examined, the first 3 have been interesting as OCD models because predictive validity has been demonstrated in each. However food restriction-induced wheel running does not correlate well on the similarity of symptoms with OCD, as body weight loss is more associated with anorexia nervosa. The 5-HT agonist model and chronic quinpirole model may be questioned on their inducing conditions; are 5-HT / dopamine dysfunction central in the pathology of OCD? Conversely, the knockout mouse model is exciting because the compulsive grooming phenotype elicited is similar to that observed in OCD. This behaviour appears to be due specifically to the knockdown of a gene that has so far not been implicated in the OCD story. However, no predictive validity has been demonstrated in this model at present. Finally, the review addresses future directions in OCD research and what possible new targets might exist.