Autism spectrum disorders constitute a significant problem in modern neurology and in neuroscience in general. At present, the incidence of such disorders is increasing, reasons for their appearance remain practically unclear, and there are no sufficiently effective treatments of these pathologies. A few animal models of autistic disorders have been developed; these models reproduce one or a few key symptoms of autism (cognitive rigidity, violations of social interactions, and qualitative disorders of communication). The respective simulations are carried out using either techniques of genetic engineering (knockout rats and mice) or early (pre- or postnatal) influences of certain environmental factors. To investigate behavioral deviations in the model animals, behavioral testing methods are used. A part of them are “classic” (e.g., the open field test, Morris water maze, T-like maze, radial maze, and Skinner’s chamber), while others have been designed specifically for models of autism. This review describes and analyzes the main methodical approaches in modeling of autism spectrum disorders in animals, and behavioral methods used in the studies of these models.