Studies of operant language therapy with autistic children have indicated that behavioral techniques may be of value in increasing the language skills of such children. A review of recent studies, however, suggests that in the absence of adequate experimental controls, misleading conclusions may be drawn about the effectiveness of treatment. It is also apparent that the effects of therapy vary according to the linguistic competence of the children involved, and to the different aspects of language ability being taught. The implications of recent findings for our understanding of the basic deficit underlying early childhood autism are discussed.