Given the heterogeneous nature of autism, identification of homogeneous subgroups is essential to furthering the understanding of the biology underlying the behaviors associated with this disorder. The availability of noninvasive, non-radioactive neuroimaging techniques and sophisticated data analytic approaches used in combination with family genetic data holds the promise of greatly improving the ability to identify these subgroups. Using these tools, Affected Sibling Pairs selected from the Autism Genetic Resource Exchange multiplex families and a sample of non-autistic control children selected from an ongoing study of normal children at UCLA will be examined. The principal goal of the work proposed in this project is to identify meaningful endophenotypes in autism. These studies will provide the basis for genetic studies to delineate specific gene-brain-behavioral pathways and lay the foundation for early detection and better intervention in autism.