I wish to comment on the phrase in the editorial by Tyrer & Craddock that in DSM-5 ‘the changes are largely cosmetic’. ^{Reference Tyrer and Craddock1} This is probably correct for most of DSM-5 but not for autism, where a new, narrow definition of autism is proposed. The broader autism phenotype is accepted by professionals in this area of study. A new study has shown that only 60% of patients meet criteria for DSM-IV autism when they are assessed using the criteria of DSM-5 autism. ^{Reference Yang, Bai-Lin and Shen2} A second error in this area in relation to DSM-5 is that an aspect of autism has been split off into a new category called social communication disorder. ICD-11 has not made this error. These changes in DSM-5 in relation to autism are radical and will lead to patients losing their diagnosis and services.