Using an example of a problem in interpretation of behavioral data, the paper asks whether behavior analysis is explanation of behavior or control of behavior. The paper argues that when two or more experimental conditions are compared, one condition is not necessarily superior to the other(s) and may not serve as an explanation of what happens in other conditions. Experimental subjects simply face different environmental conditions and behave differently in them. Successful behavior changes and management in clinic, laboratory, and school suggest that contemporary behavior analysis is behavioral engineering. But what criteria should be applied when one tries to explain behavior? Control of behavior and explanation of behavior may be separate forms of analysis with the latter incorporating elements that may not be approached directly in the manner of manipulating independent variables.