Among the normative concerns of social scientists, one of the greatest lacunae is evaluation of government performance in light of populist democratic criteria. While formal theorists have continued elaborating models of a competitive political process, and while the search for an appropriate and serviceable welfare criterion goes on, the empirical application of our existing descriptive and prescriptive insights mostly waits in abeyance. This paper offers one attempt to apply formal criteria to the evaluation task. We first briefly develop several low-level, admittedly non-complex, statements about the expected outcome of public sector activity according to the median preference model. The model’s descriptive statements then serve as evaluative and comparative criteria as we test its adequacy under several assumptions concerning democratic responsivencss. Finally we attempt to explain variations in the degree of responsiveness exhibited by ten American urban governments by reference to three aspects of political structures. The value of our approach emanates, we believe, not from any new insight into the formal interpretations and predictions provided by positive political theorists. Rather, we consider the approach a nascent attempt to match formally derived expectations to non-experimental data from both public opinion surveys and aggregate characteristics of large American cities.