Cluster analysis may offer several advantages over factor analysis in characterizing OCD heterogeneity, and this categorical approach is not limited in some of the ways Dr. Mataix-Cols et al. implied. In cluster analysis, individuals are unambiguously assigned to unique groups, whereas in factor analysis, each individual is assigned a score on all of the identified latent dimensions. Thus, the factor scores estimated for individuals may not connect the person to a specific dimension. As Dr. Mataix-Cols et al. pointed out, hoarding symptoms have emerged as a symptom dimension that predicts unresponsiveness to current pharmacotherapy and standard behavior therapy. Although there has been limited study, similar results have been reported with a cluster analysis approach in which the hoarding subgroup was less responsive to behavior treatment (3). The results of several recent cluster analyses (e.g., reference 4) suggest that complex symptom presentations can be captured with a cluster analysis approach and that resultant clusters are far from monosymptomatic.