The freedom of a doctor to treat an individual patient in the way he believes best has been markedly limited by the concept of evidence-based medicine. Clearly all would wish to practice according to the best available evidence, but it has become accepted that "evidence-based" means that which is derived from randomized, and preferably double-blind, clinical trials. The history of clinical trial development, which can be traced to the use of oranges and lemons for the treatment of scurvy in 1747, has reflected a progressive need to establish whether smaller and smaller effects of treatment are real. It has led to difficult concepts such as "equivalence" and aberrations such as "meta-analysis." An examination of evidence-based practice shows that it has usually been filtered through the opinions of experts and journal editors, and "opinion-based medicine" would be a more appropriate term. In the real world of individual patients with multiple diseases who are receiving a number of different drugs, the practice of evidence-based (or even opinion-based) medicine is extremely difficult. For each patient a judgment has to be made by the clinician of the likely balance of risks and benefits of any therapy. Good practice still requires clinical freedom for doctors.