The full participation of individuals with mental retardation in the community requires performance of newly acquired skills in novel circumstances and across the varying demands characteristic of life in the everyday world. For example, a person who has learned how to ride a bus may have to adapt to changes in scheduling, bus routes, or fares. Or an employee who has been taught by a coworker to make salads in a fast-food restaurant will need to continue to complete orders when the coworker no longer is present. Although skill generalization is an implicit educational goal, instructional strategies rarely are employed to influence the attainment of this goal (Haring & Laitinen, in press). Self-instruction is a strategy that has been effective in promoting the independent performance of people with mental retardation. Self-instruction provides individuals with the means for guiding their own behavior in novel situations not associated with training and after assistance has been withdrawn. For example, by using self-instruction, people with mental retardation have learned, among other skills, to sequence their tasks, increase their rate of production, and solve work-related problems. This chapter evaluated studies that investigated use of self-instruction among individuals with mental retardation in community settings. The focus of the review was on identification of factors relating to generalization across people, situations, tasks, and time. The combination of these factors suggests a model for promoting independent performance among individuals with mental retardation. This model combines self-instruction with teaching multiple exemplars and comprises the following six steps: (a) Select an array of examples (responses) an individual is likely to be required to perform in an environment (step 1); (b) classify responses into teaching sets based upon a functional analysis (step 2); (c) divide items of each set into responses that will serve as training examples and those that will serve as generalization probes (step 3); (d) teach trained examples using self-instruction (step 4); (e) evaluate effect of training on trained and untrained examples (i.e., generalization probes) as well as verbalized self-instructional statements (step 5); and (f) withdraw training based upon performance criteria while evaluating the effect of withdrawal (step 6). These steps represent the best practices for promoting independent performance of individuals with mental retardation in the community.