The standard formulation of the interventionist theory of causation has all the surface trappings of a conceptual analysis, but cannot succeed as such, due to circularities in its conditions for causation. Many critics who reject the theory have found this to be its fatal problem. Woodward has responded that the theory is non-reductive, and emphasized rather its usefulness for causal reasoning. I suggest a reformulation of the interventionist theory such that it can't be confused with a conceptual analysis, while it retains the utility of Woodward's original formulation. In the process I clarify the notion of "possible intervention," and trace a critical ambiguity in this notion, both to Woodward's seemingly ambiguous goals for the theory, and to its roots in two ways of talking about interventions in scientific theories of causal inference.