AIM OF REPORT. This report reviews the evidence about the effectiveness of treatments for chronic pain. While treatment of chronic pain is usually seen as an integrated service, this report concentrates on the individual interventions that constitute the service. HOW THE RESEARCH WAS CONDUCTED. Searches of databases and journals identified over 15,000 randomised studies with pain as an outcome, and many more which were not randomised. Over 150 systematic reviews relevant to chronic pain treatment were identified and their quality assessed using a simple scoring system. Systematic reviews conducted for this report were based mainly on randomised trials. The number needed to treat (NNT) was chosen as the output for the report. NNTs of 2-4 indicate effective treatments. Because NNT is treatment-specific it overcomes problems associated with highly variable placebo or control event rates in pain trials. Such variability is predominantly due to the limited numbers of patients in the clinical trials. Dichotomous outcome measures are important in synthesising information from many studies, and in deriving NNTs. Methods have been developed which allow mean information on pain relief and intensity to be converted reliably into the simple dichotomous outcome of at least 50% pain relief. RESEARCH FINDINGS. PHYSICAL INTERVENTIONS. Transcutaneous electrical nerve stimulation (TENS) has been shown not to be effective in postoperative and labour pain. In chronic pain, there is evidence that TENS effectiveness increases slowly, and that large doses need to be used. There is lack of evidence for the effectiveness of TENS in chronic pain. There is a lack of evidence for the effectiveness of relaxation. Intravenous systemic regional blockade with guanethidine has been shown to be without effect. Epidural corticosteroids are effective in the short term for back pain and sciatica. Injections of corticosteroids in or around shoulder joints for shoulder pain have been shown not to be effective. There is a lack of evidence supporting spinal cord stimulators. Case series are of poor quality and do not provide evidence of effectiveness, although at least 50% pain relief at 5 years is reported in over 50% of patients. PHARMACOLOGICAL INTERVENTIONS. Minor analgesics are important in chronic pain. NNTs were calculated for analgesics given orally for moderate or severe acute postoperative pain. The NNTs found ranged from 17 (poor) for codeine, 60 mg, to 2.5 (good) for ibuprofen, 400 mg. Anticonvulsant and antidepressant drugs are prescribed for neuropathic pains like diabetic neuropathy. NNTs are of the order of 2.5, showing them to be effective treatments. However, there are too few studies with too few patients to determine which is the best drug. Minor adverse events are common, and major adverse events occur in about 1 in 20 patients. There are no studies comparing antidepressants and anticonvulsants directly. Systemic local anaesthetic-type drugs have been shown to be effective in nerve injury pain but there is little or no evidence to support their use in migraine or cancer-related pain. Topical NSAIDs (for example, gels, creams) are effective in rheumatological conditions with an overall NNT of 3. There are too few studies to determine which is the best agent. Topical NSAIDs have few adverse events; most importantly they are without the major gastrointestinal adverse events found with oral NSAIDs, which might make them an important choice for some patients with peripheral arthritis. (ABSTRACT TRUNCATED)