This article evaluates four different ways of relating the normative side of sociology to its empirical side. Two such ways are in existence at present. The first is “dualism,” the idea that sociology provides purely scientific results to political or moral projects that are conducted on some independent normative basis. This position is commonly invoked in the idea of “value-free sociology.” The second is “monism,” the ideas that value-freedom is impossible and that sociology is inevitably value-driven, indeed perhaps that it should be openly so driven. This position is commonly invoked in the idea of “the unity of theory and practice.” These existing approaches are complemented by two that do not yet exist in practice. Both are explicitly normative in part. The first of these is a “canonical” approach, like that of the subdiscipline of political theory, in which normative inquiry within sociology would be formally recognized within the discipline and would be organized around a classical canon of normative works. The second would be a “legalist” approach, which would grow out of new genres of writing that aimed at the systematic normative evaluation of bodies of work or literatures, thus working inductively, in contrast to the canonical approach’s deductivism. The article evaluates these four positions according to four criteria: feasibility, coherence, trajectory, and open-mindedness. It concludes that the current positions (dualism and monism) are both embarrassingly weak: typically unconscious and sometimes naïve, in many cases driven by the unacknowledged – and hence uncritical - assumption that one's particular politics are in fact universally desirable. The discipline should try to create an explicit but rigorously argued normative subdiscipline, probably combining both the canonical and legalist positions.