Autism spectrum disorders (ASD) are severe neurodevelopmental disorders with marked deficits in social communication, verbal development, and behaviour. The broad phenotype and the clinical heterogeneity point to a polygenic disorder - despite high heritability among siblings. According to recent findings not only do single-rare mutations but also copy number variations and single nucleotide polymorphisms impact the ASD phenotype. Because of the scope of national and international consortia, many linkage and genome-wide association studies have evolved which elucidate candidate and susceptibility genomic regions and genes relevant for ASD. In contrast to polygenic or genetic complex models for autism, a few monogenetic forms of ASD are known to be caused by single gene defects, e.g., fragile-X syndrome. Knock-out animal models of monogenetic autism (e.g. FMRP(-/-)) or neurodegenerative disorders (e.g. MeCP2(-/-)) are often used to analyze the molecular mechanisms underlying ASD. In this review we describe the state of the art of genome analyses in ASD, the most widely used mouse models for polygenic or monogenetic forms of autism and discuss new insights gained from these analyses. The susceptibility genes so far identified seem to be involved in the proper establishment of the synaptic cleft, the secretion of surface proteins, or the overall cellular translation processes. Theses findings suggest that impacting translation-dependent processes like synaptic plasticity or cell-to-cell connectivity may lead to an ASD phenotype.