The placebo effect is often poorly understood or confused with evaluation bias or spontaneous improvement, particularly when study inclusion criteria select patients at the peak of their symptoms. Cerebral imaging studies have confirmed that the placebo effect exists, although it is now known to involve a combination of conditioned reflexes and reward anticipation. The magnitude of the placebo effect can be evaluated by randomly dividing patients into three groups, one of which receives no treatment at all; by crossover studies; or by the newly developed open-hidden study design. This last design has established that rebound effects can occur after placebo discontinuation, and other experiments have shown that anxiety is associated with a weaker placebo response. This anti-placebo effect of anxiety, similar to the nocebo effect, may involve the release of cholecystokinin. The strength of the placebo effect varies across procedures and joints. A marked placebo effect can be seen in rheumatology patients, as shown recently by two high-quality double-blind studies that found no difference between vertebroplasty and a sham procedure. Effective blinding is crucial both to obtain a strong placebo effect and to separate an intrinsic effect from a placebo effect. Beliefs of the patients and physicians regarding the active drug and the existence and strength of the placebo effect could also be usefully evaluated throughout clinical studies.