Oesophagitis is associated with Barrett's metaplasia in about 10% of individuals. The UK has one of the highest world-wide prevalences of Barrett's metaplasia, with 1% of adults having the condition, resulting in an incidence of oesophageal adenocarcinoma two to three times that seen in either Europe or North America. In addition, the conversion rate to cancer in individuals with Barrett's metaplasia in UK surveillance programmes is twice that observed in the USA (0.96% per year vs. 0.4% per year), lending further support to the notion that the UK is a high-risk region. The evidence base on what can be achieved with medical therapy to reduce the risk of dysplasia or the development of adenocarcinoma needs to be strengthened with data from randomized controlled trials, as existing data have many limitations. Patients with Barrett's metaplasia respond variably to proton pump inhibitor therapy (even high-dose therapy 'normalizes' acid reflux in only 85% of cases), and symptom control is a poor determinant of the adequacy of suppression of acid reflux. Gastro-oesophageal reflux is implicated in the pathogenesis of Barrett's metaplasia, and ex vivo and in vitro evidence suggests that its attenuation reverses proliferation and biological variables over days, and perhaps the metaplastic histology to a degree over years. The effect of proton pump inhibitor therapy on cancer risk in the long term is essentially unknown. Acid suppressant therapy or anti-reflux surgery on its own does not result in the complete regression of the metaplastic epithelium. Bile acids, present especially frequently in the refluxate of Barrett's oesophagus patients, are also likely to influence the development and persistence of metaplasia. Barrett's metaplasia is replaced by a squamous epithelium when acid reflux is well controlled and the epithelium is physically destroyed by ablation with argon plasma coagulation or photodynamic therapy. These modalities are invasive and are not likely to be useful in the routine management of patients with Barrett's oesophagus without dysplasia or cancer. Why metaplasia does not fully regress once external initiating stimuli are removed is a mystery. There is some evidence to implicate a variety of molecules, including cyclo-oxygenase-2, tumour necrosis factor-alpha, beta-catenin nuclear translocation and mitogen-activated protein kinase signalling, because they are expressed preferentially in metaplastic rather than normal or inflamed squamous oesophageal mucosa. The use of non-steroidal anti-inflammatory drugs, including aspirin, is associated with a decreased incidence of oesophageal adenocarcinoma. There is therefore a great need for randomized controlled trials to assess the outcomes of such chemopreventive therapy in patients with Barrett's metaplasia.