The concept of power derives its meanings and theoretical roles from the theories in which it is embedded. Hence, there is no one concept of power, no single understanding of power, even if these understandings stand in relation to each other. Besides the usual theoretical traditions common to the discipline of international relations and the social sciences, from rationalist to constructivist and post-structuralist approaches, there is, however, also a specificity of power being a concept used in both political theory and political practice. A critical survey of these approaches needs to cast a net wide to see both the differences and the links across these theoretical divides. Realist understandings of power are heavily impressed by political theory, especially when defining the ontology of “the political.” They are also characterized by their attempt, so far not successful, to translate practical maxims of power into a scientific theory. Liberal and structural power approaches use power as a central factor for understanding outcomes and hierarchies while generally neglecting any reference to political theory and often overloading the mere concept of power as if it were already a full-fledged theory. Finally, power has also been understood in the constitutive but often tacit processes of social recognition and identity formation, of technologies of government, and of the performativity of power categories when the latter interact with the social world, that is, the power politics that characterize the processes in which agents “make” the social world. Relating back to political practice and theory, these approaches risk repeating a realist fallacy. Whereas it is arguably correct to see power always connected to politics, not all politics is always connected or reducible to power. Seeing power not only as coercive but also productive should neither invite one to reduce all politics to it nor to turn power into the meta-physical prime mover of all things political.