The purpose of this study was to investigate systematically the feasibility of modifying the behavior of autistic children in a classroom environment. In the first experiment, eight autistic children were taught certain basic dassroom behaviors (including attending to the teacher upon command, imitation, and an elementary speaking and recognition vocabulary) that were assumed to be necessary for subsequent learning to take place in the classroom. Based on research documenting the effectiveness of one-to-one (teacher-child ratio) procedures for modifying such behaviors, these behaviors were taught in one-to-one sessions. It was, however, found that behaviors taught in a one-toone setting were not performed consistently in a classroom-sized group, or even in a group as small as two children with one teacher. Further, the children evidenced no acquisition of new behaviors in a classroom environment over a four-week period. Therefore, Experiment II introduced a treatment procedure based upon "fading in" the classroom stimulus situation from the one-to-one stimulus situation. Such treatment was highly effective in producing both a transfer in stimulus control and the acquisition of new behaviors in a kindergarten/first-grade classroom environment.Autism is a severe form of childhood psychosis characterized by the relative absence of appropriate speech, play, and social behavior, and by a relatively high frequency of bizarre, repetitive mannerisms and tantramous behaviors. A large majority of the research investigating variables in the treatment of autistic children has utilized a one-to-one teacher-child ratio lThis investigation was