The author argues that the self is a multifaceted entity that does not easily submit to clear and precise description. The aspect of self studied by most investigators is actually a subset of the cognitive and neural underpinnings of "self" and not the "self" of first-person subjectivity. The author then looks at the dominant theoretical treatment of human long-term memory-the systems approach-and examines how the construct of "self" is situated in this framework. Finally, he reviews the best-known paradigm for exploring the role of self in memory-the self-reference effect (SRE) manipulation. He argues that there is not one SRE but rather a family of related SREs that are influenced by a variety of variables and contexts. Accordingly, researchers must exercise caution when attempting to draw conclusions about the self from the results of SRE memory performance.