One of the tasks of medical anthropology is to conduct research to evaluate the efficacy of traditional health care practices. The benefits of health care may be evaluated in numerous ways, but in this article we examine only the problem of how to determine whether a therapeutic intervention changes the pathophysiology of a disease. The randomized controlled trial is acknowledged as an ideal that will rarely be attainable by medical ethnographers. Individual case studies are primarily useful for hypothesis formation. We are left then with observational studies (case series) as a feasible and useful alternative. Those presently in the anthropological literature are examined and each is found to be flawed to some extent. Future investigations can profit from what was learned in these pioneer studies by giving more attention to patient selection, treatment description, and objective measures of outcome.