This article attempts to reformulate and resuscitate the seemingly prosaic methodological task of description, which is often derided in favour of causal analysis. First, the problem of definition is addressed: what does this category of analysis (‘description’) refer to? Secondly, a taxonomy of descriptive arguments is offered, emphasizing the diversity contained within this genre of empirical analysis. Thirdly, the demise of description within political science is charted over the past century, with comparisons to other disciplines. Fourthly, it is argued that the task of description ought to be approached independently, not merely as a handmaiden of causal theories. Fifthly, the methodological difficulties of descriptive inference are addressed. Finally, fruitful research areas within the rubric of description are reviewed.