For one boy with developmental disabilities, we conducted naturalistic observations and saw that aggression occurred during instructional task presentations. Next, in an experimental analysis, we manipulated the manner in which instructions were presented. Results showed that differentially high rates of aggression were observed when the manner in which instructions were presented was manipulated. Based on the results of the assessment, a successful intervention was developed that incorporated the idiosyncratic finding observed during both the naturalistic observations and the experimental analysis. Copyright © 2004 John Wiley & Sons, Ltd.