Self-expanding metal stents provide a new option for the palliative treatment of malignant stenoses from tumors of the esophagus. Our present study provides a comparative assessment of clinical experience gained consecutively in the implementation of three stent versions in a total of 87 patients and 96 implantations. Thirty-one Wallstents (Scheider, Lusanne, Switzerland) (five coated) were implanted in 23 patients, 35 Ultraflex stents (Boston Scientific, Boston, MA, U.S.A.) (uncoated) in another 34 patients, and 30 Gianturco-Z stents (Cook, Winston-Salem, NC, U.S.A.) (all coated) in a group of 30 patients. In the three patient groups there were no significant differences as to the degree of dysphagia, number of pretreatments, length of the tumor stenosis, tumor location, or histological classification. Seven patients who had been treated with Gianturco-Z stents presented with an esophagorespiratory fistula. Technically, all 96 implantation procedures were successful. Complete sealing of the fistulas was verified by radiography in all patients who had developed fistulas. Severe early complications in the form of stent migration were encountered in only three of 96 implantations (3%). Within the early period after stent placement in five patients of the Wallstent group (22%), 13 patients of the Ultraflex group (37%), and three patients of the Gianturco group (10%), retreatments were necessary due to stent dislocation and/or insufficient stent expansion. The degree of dysphagia improved distinctly and with a comparable development in all three patient groups. The rate of reintervention (percentage of patients) due to major and minor problems in the follow-up period amounted to 43% (Wallstents), 35% (Ultraflex stents), and 21% (Gianturco-Z stents). In view of the low number of complications (3%) in the early stage of implantation, self-expanding metal stents provide an improved approach for palliative therapy of malignant stenoses of the esophagus. Nevertheless, further technological improvements are necessary to reduce the great frequency of unavoidable reinterventions (20-43%) in the follow-up period.