<p>Philosophers of science frequently appeal to “complexity” to justify scientific pluralism—the view that some domains require multiple, perhaps non-integrable, models or explanations. This article argues that such appeals often rely on an unexamined metaphysical interpretation of complexity, according to which complexity is an intrinsic property of the world capable of grounding normative methodological prescriptions. I show that this interpretation is both unjustified and unproductive. Through critical analyses of Mitchell’s integrative pluralism, Waters’s “no-structure” argument, and Longino’s account of human behavior, I show that their supposed metaphysical claims about complex systems are in fact epistemic claims in disguise. Once this is acknowledged, metaphysical complexity cannot sustain the strong methodological conclusions pluralists often draw from it. In place of metaphysical foundations, I propose a pragmatist account of complexity. Drawing on Wittgenstein’s notion of family resemblance and Neurath’s concept of <i>Ballungen</i>, I argue that complexity functions as a flexible, non-essentialist cluster concept whose meaning is stabilized through overlapping similarities and paradigm cases, rather than necessary and sufficient conditions. This view preserves the heuristic and organizational role of complexity in scientific inquiry while avoiding unwarranted ontological commitments. Complexity remains philosophically and scientifically useful not because it designates a natural kind, but because it coordinates diverse investigative strategies and supports context-sensitive forms of pluralism. A genuinely pragmatist pluralism, I conclude, requires abandoning the search for metaphysical grounding in favor of a reflexive, non-foundationalist stance.</p>