Animal models have shown progressive development and have undoubtedly proven their supportive value in OCD research. Thus, various animal models have confirmed the importance of the 5-HT [72-74] and dopamine systems [104,111] in the neurobiology and treatment of OCD. Given the neurochemical, emotional, and cognitive complexity of the disorder, how-ever, animal models are being used to investigate more and more complicated neurochemical and behavioral theories purported to underlie OCD. The lever-press model, for example, has implicated deficient response feed-back in a neural system that regulates operant behavior [74]. Studies on stereotypic movement disorder [89] have opened a new avenue of investigation into the neurobiology of stereotypy that may be applicable to more complex syndromes such as OCD. Models that have focused on specific neuropsychologic aspects of OCD such as reward [74], displacement behavior[63,101], perseveration and indecisiveness [73,102], and spontaneous stereotypy [90,94] are important in their attempt to unify the diverse behavioral manifestations of this disorder. It is clear that for a deeper, more holistic understanding of OCD, multiple animal models will be needed to allow investigation of the various aspects of the disorder and to provide convergent validation of the research findings. The heterogeneous nature of OCD, the various subtypes that exist within the disorder, and the range of obsessive-compulsive spectrum disorders suggest that particular questions regarding OCD may be addressed best by us-ing a particular ethologic model, whereas other questions might require a pharmacologic model or a combination of both for meaningful results[62,115]. Genetic models will be extremely useful for studying the genetics of pathologic behavior and for relating these findings to neuroanatomic and neurochemical changes in the model (eg, DICT-7 mice as a model for Tourette's syndrome and OCD). Neither ethologic nor pharmacologic models, however, can assess whether the "compulsive" behavior is a response to an "obsessive" anxiety or fear. Perhaps the symptoms seen in patients who have OCD, which may be exacerbated by everyday stress, are analogous to displacement behaviors in animals and also reflect some form of anxiety or stress [98]. In this regard, the bank vole model [116]has provided evidence that previously developed stereotypies increase markedly after acute stress and argues that healthy individuals "habituate" to everyday stress, whereas patients who have OCD do not. Interindividual variation in behavioral response and attempts to replicate studies in different laboratories often is the nemesis of the behavioral scientist. Small within- and between-subject variability is usually desirable, how-ever, because there are cases in which the study of the variability of the model could lead to a better understanding of the disorder. Variability can-not always be considered an error; it is possible that previously disregarded neuronal systems may have a place in the observed variation and, indeed, in the pathophysiology of OCD. In this regard, SRIs are not always effective for OCD [6,29,30] such that a lack of effect in a model may reflect an un-known neurobiological basis for compulsive behavior in a sub-group of SRI refractory patients. Similarly, separating the afflicted (ie, working with animals that show greater behavioral change in a model and/or after drug treatment) would have distinct benefits. To increase successful implementation of an ethologic animal model, especially when reinforcement models or signal attenuation models are used,the laboratory must be equipped with the essential behavioral testing apparatus as well as the operant chambers/rooms in which to conduct the train-ing and data collection. Quantification of certain stereotypy behaviors also requires experienced or trained observers. An illustration of the difficulty in measuring behavioral changes is that in the rewarded alternation model,a good response to behavioral treatment (alternation training) may lead to a floor effect [73] which, after successful drug treatment of the animal,produces no residual persistence (ie, measurable behavioral change) on which a drug treatment can be tested. Clearly, the choice of ethologic, pharmacologic, or genetic models should be considered carefully. A well-validated model may quell many of the limitations and considerations described previously. Noninvasive neuroimaging(eg, the use of small-animal single-photon emission CT) to explore the neuroanatomic basis of OCD offers an exciting future challenge, especially if combined with pharmacologic or ethologic models, and could confirm or ex-tend knowledge of the neuroanatomy of OCD. Although studies to investigate further the interactive role of 5-HT, dopamine, GABA, and glutamate are still needed, the role of neuroactive peptides such as cholecystokinin, corticotrophin-releasing factor, neuropeptide Y, tachykinins (ie, substance P),and natriuretic peptides in OCD should also be considered. Genetically engineered animal models will become increasingly valuable in combination with new technologies such as gene-chip microarrays, RNA interference, and advanced proteomics that will help further the understanding of OCD. Animal models of OCD are poised to play a vital role in extending the knowledge of the disorder now and in the future.