The recent technical improvements for the study of the cytogenetic basis of disease have led to the identification of many microdeletion and microduplication syndromes. De novo copy number variants (CNVs) are seen more often than expected in autistic patients, and rare chromosomal aberrations are known to account for a small fraction of schizophrenia and bipolar disorder. Recurrent spontaneous mutations at multiple sites across the genome may be a prime cause of these disorders, and the reduced fertility associated particularly with autism and schizophrenia may, through selection pressure, maintain these variants at very low frequency. CNVs are large enough to be identified using cost efficient genome-wide search techniques, such as oligonucleotide arrays. In contrast, rare DNA mutations might confer a large proportion of the overall genetic risk, but cannot be easily identified at the present time. Genetic epidemiology suggests that autism, bipolar disorder, and schizophrenia are highly heritable and share susceptibility genes. However, the results of dense SNP array-based genome wide association scans for these disorders have been disappointing, as no high frequency risk alleles have been identified. The PsychCNVs consortium will apply oligonucleotide arrays for the large-scale interrogation of CNV variation in the human genome, focusing on people with autism and childhood onset schizophrenia and bipolar disorder, where CNVs are likely to be more common. This project follows on logically from our hypothesis-independent genome-wide SNP search for common variants, we now intend to systematically search for large, rare CNVs conferring high risk. The genetic risk conferred by CNVs will be estimated by genotyping samples from 2800 patients, to identify de novo or recent spontaneous mutations. Until low cost whole genome sequencing becomes available this approach offers the best current hope to identify causative genomic variants for these disorders.