We examine the relationship between functional brain networks and behaviour in individuals with autism from the Autism-Brain-Imaging-Data-Exchange. We find a difference in attention network strength between groups (autism vs. control) using an a priori defined network for high-attention . In addition, connectome based predictive modeling (CPM) successfully predicted inattention and communication scores. Finally, on a test group of autistic patients (not used in the CPM), we found network strengths to be more similar to individuals used the CPM with autism than controls. Together, these results indicate that connectivity may prove to be a valuable tool in the diagnosis and treatment of autistic individuals.