The aim of this study was to examine whether a low cost self-generated feedback procedure was sufficient for generalising and maintaining staff performance. Two staff members were trained to implement a communication skills program with a client with severe closed head injuries. Staff were first trained to perform competently via written instructions, videotaped models, discussion, and on-the-job supervisory feedback. Posttraining conditions consisted of staff continuing with recording and graphing gains in client skills, or rating their own performance (i.e., they continued with self-generated feedback). A multiple baseline design across client skills and staff was used to observe generalised effects on staff performance (prompts and positive consequences) and client skills. Results show that after staff attained competency, and without further supervisory feedback, their performance generalised to topographically similar performance, and the staff member who consistently participated in the study maintained her performance for 16 weeks.