Outcome research has shown that early and intensive behavioural intervention (ABA) may improve intellectual, language and adaptive functioning in children with autism. However, research has also indicated that not all ABA provisions are equally effective. Therefore, it may be beneficial to describe the key variables that are common to programmes which have been empirically validated through outcome research. This paper describes a research-based service model which has been empirically validated. Important components include treatment in the child's natural environment, intensive intervention, treatment based on applied behaviour analysis, staff training and management, parental involvement, evaluation of progress, and research-based provision. (Contains 1 figure.)