Functional Analysis is an important component of creating a treatment plan for individuals with autism. A concern of functional analysis is that it might not match the natural environment in which the problem behavior occurs. Dimensions such as varying reinforcement duration and session length may affect response rates during functional analyses in natural settings. This study investigated if varying reinforcement duration and session length during functional analyses would affect the identification of the function of problem behavior across assessors, and the intervention designed based on the functional analyses would be effective in reducing stereotypic behavior and teaching communicative skills of an adolescent with autism at home. An alternating treatment design was used to conduct functional analyses and an ABA design with generalization probes was used to test the effectiveness the function-based intervention. Results indicated that the participant's stereotypic behavior was maintained by automotive reinforcement. The intervention developed based on the function of the stereotypic behavior was found to be effective in reducing the stereotypy and teaching communicative behavior.