The dominant research tradition in psychology, including much of cognitive-behaviour therapy, requires that large numbers of subjects be allocated randomly to form treatment groups. Treatment effects typically are assessed by testing a null hypothesis about group mean differences. This paradigm seriously thwarts the development of a science of individual behaviour, inhibits the implementation of the scientist–practitioner model, stifles innovation, impairs accountability, and precludes the scientific investigation of the exceptional or novel case. Single-case research designs, first systematically expounded by Sidman (1960), make it possible to draw scientifically valid conclusions from the investigation and treatment of individuals. Criticisms of the between-groups research paradigm are summarised, the key elements of single-case designs are outlined, and their consistent adoption strongly recommended.