The clinical presentation of obsessive-compulsive disorder (OCD) varies not only across patients but over the course of the disorder. This diversity indicates that OCD is a heterogeneous disorder, which may have an important impact on psychopathological, longitudinal, genetic, and treatment research. To better understand OCD heterogeneity, more homogeneous phenotypic descriptions are necessary to delimiting clinically meaningful subgroups of patients. Besides phenotypic descriptions, another method of delimiting OCD patient subgroups includes the search for endophenotypes (extended phenotypes) based on neurophysiological, immunological, genetic, neuropsychological, or neuroanatomic (neuroimaging) paradigms. This article will describe some strategies that deal with OCD heterogeneity, including the identification of more homogeneous phenotypical categories, an improved understanding of obsessive-compulsive symptom dimensions and how to use them as quantitative traits, and broadening the diagnostic boundaries of OCD to include other related conditions. The relevance and limitations of each approach are also discussed. Since the etiological mechanisms associated with the expressions of OCD are unknown, there is probably not one but several heuristic strategies to search for more homogeneous OCD subgroup, that combined may provide the most fruitful results.