Larson et al ^{Reference Larson, Wagner, Jones, Tantam, Meng-Chuan and Baron-Cohen1} report on a major study on psychosis in autism, which is an important topic. ^{Reference Fitzgerald2} They point out that their sample is non-representative, but then use the Autism Diagnostic Interview-Revised (ADI-R) and Autism Diagnostic Observation Schedule (ADOS) for clinical diagnosis. This vastly increases the non-representativeness of the sample unnecessarily and takes it a very long way away from autism in the general population. The criteria they used to define autism are very narrow concepts of the disorder. Clinical diagnoses based on this narrow view tell us very little about autism as seen in routine clinical practice, where professionals throughout the world now accept that the broader autism phenotype. I see many parents who come to me in great distress knowing that their child has autism and that the school also observed this, but having been told that their child did not have autism according to the ADI-R. This instrument is not appropriate to making a sole diagnosis of autism in clinical practice. It not uncommonly misses high-functioning autism. In addition, Ventola et al ^{Reference Ventola, Kleinman, Pandey, Barton, Allen and Greene3} have shown that the ADI-R was significantly ‘under-diagnosing toddlers’. How biased and unrepresentative the patients in this survey can be seen by Professor Gillian Baird's work on autism in the general population. ^{Reference Baird, Simonoff, Pickles, Chandler, Loucast Meldrum and Charman4} Indeed, using these narrow criteria gives a prevalence of autism of 25 per 10 000. When you use the broader autism spectrum, you get a truer rate of 116 per 10 000. One of the problems also is that the National Institute for Health and Care Excellence (NICE) guidelines on the diagnosis of autism, ⁵ which are accepted throughout the world, are not followed. These state that there is no specific instrument recommended for diagnosis of autism and that identification depends on a clinical diagnosis by an experienced clinician. Dorothy Bishop, Professor of Developmental Neuropsychology at the University of Cambridge, told Adam Feinstein that, ‘If it could be shown that there were real benefits in accuracy of diagnosis from adopting this lengthy procedure, then I'd be happy to say: “Okay”. But the originators of the instrument have never demonstrated [this] – it is really more an article of faith with them’. ^{Reference Feinstein6} Feinstein also reports that, at the prestigious International Meeting for Autism in London in 2009, senior autism researchers ‘lambasted’ these narrow instruments ‘for missing many cases of autism’.