Does it work? This question lies at the very heart of the kinds of controversies that have surrounded complementary and alternative medicines (such as herbal medicine) in recent decades. In this article, I argue that medical anthropology has played a pivotal and largely overlooked role in taking the sham out of the placebo effect with important implications for what it means to say a therapy or drug `works'. If pharmacologists and clinicians have corporeally located the concept of efficacy in terms of bio-availability, pharmacodynamics and pharmacokinetics, and herbalists in terms of a herbal revitalizing of the body's own vis medicatrix naturae, from the early 20th century onwards medical anthropologists (especially those who became interested in the `savage mind') have built up an equally rigorous theory of symbolic efficacy in terms of narratives, symbols and a kind of cognitive homeostasis. It was precisely as a mediating link between the somatic and the symbolic that I suggest a decriminalized placebo effect (as opposed to suggestion) could emerge in the middle of the 20th century. Taking the example of St John's Wort, I go on to show how notions of symbolic efficacy, spillover placebo efficacy and bio-efficacy co-circulate in recent attempts by herbalists, clinicians and pharmacologists to address the question of whether or not this herbal remedy `works' in the treatment of depression.