The US outcomes movement assumes, and sometimes argues, the primacy for medical practice of probabilistic knowledge derived from methodologically rigorous statistical studies. 'Evidence-based medicine,' then, is considered a course of clinical medicine prescribed by such research. Implementation of evidence-based medicine as recently been uneven in the US, manifesting not only the expected 'obstacles to implementation' but several theoretical weakness of the applied science model of medical care. Outcomes researchers claim to provide certainty - certainty of what is probable, as it turns out - in a world of clinical uncertainty. This paper argues that outcomes research actually exacerbates the inferential uncertainty of practising physicians who would use knowledge for practice. Two quandaries are discussed: whether to privilege rigorous or relevant research, and whether to privilege universal or local knowledge. In each case, the logic of 'evidence-based medicine' is insufficient to resolve the quandary and would seem to support conflicting resolutions. Recent developments in US health policy are cited as manifestations of these quandaries. Finally, the reader is asked not to disregard the political implications of the outcomes movement.