The results obtained for the treatment of 59 patients diagnosed with Barrett's oesophagus, randomized to receive medical treatment (n = 27) or antireflux surgery (n = 32) were assessed prospectively. Median follow-up for the patients undergoing medical treatment was 4 (range 1-11) years and for patients undergoing surgical treatment 5 (range 1-11) years. Satisfactory symptomatic control (excellent to good results) was achieved in 24 patients after medical therapy and in 29 after antireflux surgery. The proportion of patients with persistent inflammatory lesions (54 per cent) and persistent or recurrent stenosis (47 per cent) was significantly higher after conservative treatment than after surgery (5 and 15 per cent, respectively). A decrease in the length of the segment of columnar mucosa was observed in eight of the patients who underwent antireflux surgery, and in only two of those given medical therapy. Conversely, an upward progression of the columnar lining was more frequent in the latter group (11 versus three). Mild dysplasia was observed in five patients, all from the group undergoing medical treatment. Severe dysplasia was detected in two patients, one undergoing medical treatment and the other following surgical therapy, in whom an antireflux procedure had failed previously. Both patients underwent oesophageal resection, with confirmation of a carcinoma in situ. The patients in whom antireflux surgery proved effective showed no dysplastic change or progression to adenocarcinoma. These results, despite the small number of patients and methodological limitations, question the systematic conservative approach in the initial management of patients with Barrett's oesophagus.