The question of whether sociology progresses, and how, has been an issue within sociology itself. In this chapter, the reasons for this are explored. The first set relates to the status of ‘theories’ in sociology, which, despite historical aspirations to universality, are not predictive systems that generate puzzles but second-order definitions and ideal types, which abstract over intelligible world of the subjects. They can loosely be said to progress in the sense of providing new ways of framing in response to generically defined concerns, such as the stability of elites, and novel social situations. The second set relates to quantitative models. Examples from causal modelling, a basic form of statistical modelling, are discussed to show how this same problem bears on them. They depend on assumptions about which correlations can be plausibly regarded as causal, and which are outside the system and can be ignored. As the social world changes, these assumptions gain or lose plausibility, and the models themselves lose applicability and predictive power. Change here amounts to providing a better fit to novel situations. But the models are purpose relative and the aims, and therefore progress, are externally defined, typically by changing normative or policy concerns.