Autism, at its most extreme, is a severe neurodevelopmental disorder, and recent studies have indicated that autism spectrum disorders are considerably more common than previously supposed. However, although one of the most heritable neuropsychiatric syndromes, autism has so far eluded attempts to discover its genetic origins in the majority of cases. Several whole-genome scans for autism-susceptibility loci have identified specific chromosomal regions, but the results have been inconclusive and fine mapping and association studies have failed to identify the underlying genes. Recent advances in knowledge from the Human Genome and HapMap Projects, and progress in technology and bioinformatic resources, have aided study design and made data generation more efficient and cost-effective. Broadening horizons about the landscape of structural genetic variation and the field of epigenetics are indicating new possible mechanisms underlying autism aetiology, while endophenotypes are being used in an attempt to break down the complexity of the syndrome and refine genetic data. Although the genetic variants underlying idiopathic autism have proven elusive so far, the future for this field looks promising.