Autism is a heterogeneous disorder with considerable clinical diversity, aetiological heterogeneity, and multiple accompanying disorders. Conceptualization of the core autistic symptoms has evolved from the single criteria of DSM III [1] to a combination of multiple criteria in the traditional dimensions of disturbances in social relation and communication, restricted interests and repetitive behaviours of more recent classification systems. With time, empiric research leads to changes in the combination of dimensions and specific criteria: from 3 dimension with a combination of 12 criteria for autistic disorder of the DSMIV [2], the new DSM-5 [3] defines the two dimensions (social relation/communication and restricted interests/ repetitive behaviours), with 7 criteria for the ‘‘Autism spectrum disorder’’ (ASD). From five previously independent disorders grouped as pervasive developmental disorders (PDD) in DSM-IV, DSM-5, still provisional at the time this editorial was written, defines a single Autism spectrum disorder, and creates the new diagnostic category of Social Communication Deficit describing patients with social communication impairment without significant restricted interest or repetitive behaviours. To manage autism phenotypic complexity and to assist the diagnostic and severity assessment, significant efforts have been devoted in developing reliable instruments able to accurately define and quantify specific autistic symptoms domains as well as behavioural and historical information from different sources. In this Issue, Falkmer and co-workers [4] systematically reviewed the accuracy, reliability, validity and utility of a series of accurately designed instruments and tools, including four observational schedules, six parent/carer interviews, two screening tools and four Asperger syndrome specific instruments, all supported by adequate validity and reliability data. They report that combined ADOS and ADI-R show the best correct classification rate for both DSM-IV autistic disorder and the more broad ‘‘Autism Spectrum Disorder’’ (similar to DSM-IV PDD-NOS); different versions of these instruments have been shown to be able to capture information needed to formulate the diagnosis in different ages of life, including very young children, and in patients with very different developmental levels. Many other instruments are analysed in the paper, showing that they were either lacking important information on sensitivity and specificity or did not have a large evidence-base from sufficient independent studies. Others, such as the CARS, are based mainly on clinical judgement rather than on clearly operationalized procedures. All the instruments analysed in the paper were developed using DSM-IV constructs: interpretation of these results should consider recent papers on the use of the same diagnostic instruments in evaluating the impact of the DSM-5 criteria for the diagnosis of the ‘‘new’’ ASD and the need of accurate evaluation of their quantitative algorithm. Indeed, preliminary finding suggests that a significant portion of patients currently diagnosed by DSM-IV-TR criteria may not be diagnosed by the new criteria, raising the concerns over the danger that, in many countries, these patients may lose access to services, causing significant suffering to them and their families [5]. In an epidemiological study, Mattila et al. [6] found that using the new DSM-5 criteria, only 46 % of the 26 children with DSM-IV PDD (all with full scale IQ higher than 50) were identified as ASD; similar results (only 60.6 % of A. Zuddas (&) Department of Biomedical Sciences, University of Cagliari and Child Neuropsychiatry, Cagliari University Hospital, Via Ospedale 119, 09124 Cagliari, Italy e-mail: azuddas@unica.it