Much psychological research employs the categories of extremism and moderation as categories of analysis (e.g. to identify the psychological bases for, and consequences of, holding certain positions). This paper argues these categorizations inevitably reflect one's values and taken-for-granted assumptions about social reality and that their use as analytic categories limits our ability to explore what is really important: social actors' own constructions of social reality. In turn we argue that if we are to focus on this latter, there may be merit in exploring how social actors themselves use the categories of moderation and extremism to construct their own terms of reference. That is we propose to re-conceptualize the categories of moderation and extremism as categories of practice rather than analysis. The utility of this approach is illustrated with qualitative data. We argue that these data illustrate the importance of respecting social actors' own constructions of social reality (rather than imposing our own). Moreover, we argue that categories of moderation and extremism may be employed by social actors in diverse ways to construct different terms of reference and so recruit support for different identity-related projects.