Objective To identify prognostic factors for the long-term success of transcervical resection of endometrium (TCRE). Design A retrospective analysis. Subjects 301 patients, who chose endometrial resection rather than hysterectomy and who underwent their first TCRE between 17 October 1989 and 31 December 1994.Interventions Endometrial resection carried out by one surgeon (D.W.S.). Main outcome measures Patient satisfaction, degree of bleeding and degree of pain, 1-6 years following TCRE. The procedure was considered to have failed if hysterectomy was necessary later. Results 301 patients underwent a total of 329 procedures (TCRE) and were followed up for 1-6 years (mean 2.8). Complications were few and there was no death or major morbidity. About 80% of patients were satisfied with the procedure and had either amenorrhoea (34%), scanty periods (33%) or moderate/acceptable periods (14%). The incidence of hysterectomy because of the failure of this procedure was 15%. The success rate was significantly higher in the older age group (>40 years); in those without previous moderate or severe dysmenorrhoea; where histological examination of the endometrial chippings did not show any abnormality; where further surgery such as diagnostic hysteroscopy, laparoscopy or repeat resection (repeat TCRE) was not required, and with increasing operator experience. Conclusion TCRE should remain a useful option in the management of women with menorrhagia. Greater awareness of the prognostic factors identified in this study will improve preoperative counselling and the appropriate selection of patients for TCRE.