The two-way interplay between research and clinical practice in relation to autism is reviewed with respect to: (1) diagnosis and syndrome delineation; (2) the nature of the disorder; (3) intervention studies; and (4) aetiology, as manifest during four time periods; (a) the 1950s and 1960s; (b) the 1970s into the mid 1980s; (c) the late 1980s and early 1990s; and (d) the late 1990s. It is concluded that clinical practice has changed out of all recognition during the last 50 years and that research findings have been crucial in bringing about that change. It has not, however, been a one-way traffic. Many key advances were prompted by astute clinical observations and some extravagant research claims were given a more balanced perspective through the light of clinical experience. Crucial research and clinical tasks remain but the means to meet them are there if the opportunities are taken and attention is paid to the lessons of the past.