The article in this issue by Hoeft et al,1 using voxel-based neuroimaging analyses, shows significant and distinct differences between the brains of preschool children with fragile X syndrome (FXS) and idiopathic autism as compared with controls. The authors found that brain regions implicated in social cognition, ie, frontal and temporal gray and white matter regions, were aberrant in preschool boys diagnosed with both syndromes as compared with controls; however, the differences were in opposite directions in these brain regions. Moreover, no brain differences were found between those with FXS with and without autism. The authors conclude that these distinct neuroanatomical patterns underscore the neurobiological heterogeneity of idiopathic autism diagnosed using DSM-IV clinical criteria and Autism Diagnostic Interview, Revised and/or Autism Diagnostic Observation Schedule criteria, where autistic symptoms are viewed as a final common pathway of brain dysfunction. This conclusion raises a critical question for clinical neuroscience. How should we consider the relationship of brain and behavior in the diagnosis of autism when the comparison is between a known neurogenetic disorder and a behaviorally defined diagnosis of unknown etiology? More specifically, should conditions with a known neurobiology and behavioral phenotype be used as models for a DSM-IV clinical diagnosis, especially for idiopathic autism or autism spectrum disorder?