Autism spectrum disorders (ASD) are among the most severe developmental psychiatric disorders known today, characterized by impairments in communication and social interaction and stereotyped behaviors. However, no specific treatments for ASD are as yet available. By enabling selective genetic, neural, and pharmacological manipulations, animal studies are essential in ASD research. They make it possible to dissect the role of genetic and environmental factors in the pathogenesis of the disease, circumventing the many confounding variables present in human studies. Furthermore, they make it possible to unravel the relationships between altered brain function in ASD and behavior, and are essential to test new pharmacological options and their side-effects. Here, we first discuss the concepts of construct, face, and predictive validity in rodent models of ASD. Then, we discuss how ASD-relevant behavioral phenotypes can be mimicked in rodents. Finally, we provide examples of environmental and genetic rodent models widely used and validated in ASD research. We conclude that, although no animal model can capture, at once, all the molecular, cellular, and behavioral features of ASD, a useful approach is to focus on specific autism-relevant behavioral features to study their neural underpinnings. This approach has greatly contributed to our understanding of this disease, and is useful in identifying new therapeutic targets.