DSM-5 draft criteria, revised and posted January 2011, aim to simplify the diagnostic criteria used to define the range of autistic disorders subsumed under the heading Pervasive Developmental Disorders in DSM-IV-TR. The proposal to combine Asperger disorder, pervasive developmental disorder–not otherwise specified (PDD-NOS), and autistic disorder into one new category of ASD has major implications for recognition and service provision. One critical assumption underlying the revision is that autism is a heterogeneous condition, with a considerable range of clinical severity. The range of associated intellectual disability does not wholly account for that variance in severity. Evidence for the validity of conventional subtypes, defined by DSM-IV-TR is inconsistent. A second assumption is that diagnoses within the autism spectrum may be best served by combining categorical with dimensional approaches to summarizing symptom profiles. A third assumption is that diagnostic exclusion rules should be relaxed, insofar as it makes clinical sense separately to record comorbidities. In this Panel we aim to challenge the evidence both for and against these assumptions, bringing together recent research based upon epidemiological data on autism traits in the general population and from clinically identified populations in the USA and Europe.