Abstract
Single case design is more appropriate than randomised controlled trials for studying treatment effects. Four single case designs are described and evaluated: the use of acontrol task, multiplebaselines,item-specific, andcross-overdesigns. It is suggested that evidence of qualitative changes in language can also be indicative of a treatment-specific effect. There are now published studies which illustrate all the described designs, and which describe treatment for a whole range of different language impairments. There are as yet rather few replications of treatment studies. It is suggested that single case efficacy designs are appropriate for clinical as well as research use.