Drash and Tudor describe six sets of reinforcement contingencies which may be present in the environments of some children eventually diagnosed with autism and suggest that these contingencies account for the etiology of "autistic" behaviors. Nevertheless, merely observing such contingencies in the environments of these children is insufficient to establish a positive correlation between the contingencies and "autistic" behaviors, let alone a causal relationship. To demonstrate a positive correlation, it is necessary to present evidence that the relevant contingencies are present more often in the environments of children exhibiting these behaviors than in the environments of children not exhibiting these behaviors. This condition has not been met, since no evidence to the effect that such contingencies are absent in the environments of typical children or children with disabilities other than autism has been presented. In fact, the opposite appears to be true, as is argued in the present commentary. It appears that Drash and Tudor's account of autism etiology is incomplete in that it neglects the role of unlearned differences between children and their possible interactions with the social environment in shaping "autistic" behaviors. Despite the misconception held by some that behavior analysis is a radically environmental approach, unlearned differences may be discussed within a behavioral framework. A "completely behavioral" account may discuss such differences in terms of susceptibility to reinforcement or punishment, speed of conditioning, or other unlearned characteristics which are potentially testable.