Autism spectrum disorder (ASD) represents a major mental health problem with estimates of prevalence ranging from 1/500 to 1/2000. While generally recognized as developmental in origin, little to nothing is certain about its etiology. Currently, diagnosis is made on the basis of a variety of early developmental delays and/or regressions in behavior. There are no universally agreed upon changes in brain structure or cell composition. No biomarkers of any type are available to aid or confirm the clinical diagnosis. In addition, while estimates of the heritability of the condition range from 60 to 90%, as of this writing no disease gene has been unequivocally identified. The prevalence of autism is three- to four-fold higher in males than in females, but the reason for this sexual dimorphism is unknown. In light of all of these ambiguities, a proposal to discuss potential animal models may seem the heart of madness. However, parsing autism into its individual genetic, behavioral, and neurobiological components has already facilitated a 'conversation' between the human disease and the neuropathology and biochemistry underlying the disorder. Building on these results, it should be possible to not just replicate one aspect of autism but to connect the developmental abnormalities underlying the ultimate behavioral phenotype. A reciprocal conversation such as this, wherein the human disease informs on how to make a better animal model and the animal model teaches of the biology causal to autism, would be highly beneficial.