Background: The paper reports on a laboratory investigation of the effects of self-modelling on stuttering rate in adolescents and adults. Self-modelling refers to a therapeutic or training method, usually involving videotape, that uses exposure to oneself performing selected error-free behaviours as the conduit for promoting behaviour change. Aims: To investigate self-modelling in single-subject experiments to determine whether any reductions in stuttering could be directly attributable to watching self-modelling videotapes and to ascertain whether instruction to focus attention on the target behaviour, namely stutter-free speech, was necessary for experimental effects. Methods & Procedures: A single-subject withdrawal design was employed with two adult men and an adolescent boy. Speaking sessions during all phases of the study were of 3-min duration. During the B Phase, subjects watched one of their self-modelling videotapes, which had been edited to remove all stuttered speech, before each speaking session. In the B+C Phases, subjects continued to watch the self-modelling videotapes before each speaking session and, in addition, were instructed to attempt to speak during the session as they had spoken on the videotape. Outcomes & Results: Stuttering reduced under self-modelling conditions for one of the three subjects. This effect was observed during the B+C Phases, which included instruction to attend to the target behaviour. Conclusions: The study provides laboratory evidence that self-modelling can ameliorate stuttering. The clinical implications of this finding are that treatments for adults may well benefit from the inclusion of self-modelling procedures, and self-modelling procedures may form a part of relapse prevention and management strategies.