Evidence of effectiveness is increasingly used to determine which health technologies are incorporated into public health provision. Acupuncture is a popular therapy that has been shown to be superior to placebo in the treatment of nausea and dental pain, and promising for migraine and osteoarthritis of the knee. For many other conditions, such as chronic pain, in which acupuncture is often used, the evidence is either insufficient or negative. Misleading results may occur for a number of reasons. False negative results may arise from inadequate treatment schedules and inappropriate control interventions. This consensus document considers these issues with the aim of improving the design of efficacy trials of acupuncture in order that they are more likely to be conclusive and more meaningfully interpreted. Clinical trials of acupuncture must use an optimal form of treatment; this can be defined by examining standard texts, by surveying and consulting experts. There are a great many variables in treatment (such as point selection, form of stimulation) all of which need to be addressed in designing and reporting clinical trials. The control procedure is determined by the precise research question that is being addressed. For efficacy studies, in which the question is whether acupuncture has specific effects (i.e. is superior to placebo), sham forms of acupuncture appear the most appropriate method of controlling for needle penetration. A recent development of blunted, telescopic needles may represent a major advance. Such procedures may produce a therapeutic response so should preferably be recorded as 'sham' procedures rather than true 'placebo' controls. Blinding in clinical trials is an accepted means of reducing bias. Patient blinding in acupuncture studies can be achieved by sham procedures and its success should be measured. While practitioner blinding is difficult, though not impossible, blinding of the observer and the analyst should be considered as the ideal for all studies. A number of recommendations are made which aim to improve the quality of sham-controlled acupuncture studies.