Autism is currently conceptualised as a unitary disorder, in which social-communication impairments are found alongside repetitive interests, behaviours and activities (RIBAs). This relies upon the validity of the assumption that social-communication impairments and RIBAs co-occur at an above chance level as a result of sharing underlying causes. In the current review it is argued that the evidence for this assumption is scarce: the very great majority of RIBA research has not been intended for or suited to its examination. In fact only three studies are fit to address directly the question of the relationship between social-communication impairment and RIBAs, and these contradict each other. In consequence, further relevant evidence was sought in the behavioural and genetic literature. This approach suggested that the correlation between social-communication impairments and RIBAs has been exaggerated in the current consensus about the autism syndrome, and that these aspects of autism may well share largely independent underlying causes. Some clinical and research implications are discussed.