DescriptionChildhood psychopathology refers to a heterogeneous set of psychological conditions that negatively influence functioning. To improve treatment, effort has been directed at defining, and categorizing disorders. The Diagnostic and Statistical Manual, the primary source for diagnostic information in the US, updates diagnostic criteria to parallel research and clinical advances. Nevertheless, much symptom overlap remains across conditions, complicating diagnosis and slowing research progress. Latent class analysis (LCA), a person-centered analytic approach, was used to explore new diagnostic groupings based on primary and comorbid diagnostic data from children with a diagnosis of Autistic Disorder or Asperger's Syndrome (ASD) (n = 76) or Bipolar Disorder (BPD) (n = 36), compared to 27 controls. LCA was expected to identify a subset of children with high comorbidity who would demonstrate distinct neuroanatomical and behavioral profiles. Comparison of the temporal cortex, amygdala, or hippocampus volumes between the diagnostic groups, and between the derived clinical latent classes, revealed no significant differences. The diagnostic groups were different on several problem behavior subscales, as were the latent classes. All clinical groups had more behavioral problems compared to controls. Although results did not support the use of comorbid information to improve diagnostic profiles, large within-group variances in the primary diagnostic groups supported the need to improve differential diagnoses. The DSM-IV categorical classification system is limited in its ability to characterize 'comorbid' symptomology. In the DSM-V, inclusion of a dimensional component and 'cross cutting' symptoms would provide clinicians with a useful way to differentiate disorders and evaluate symptom severity.