Résumé : Objective: The ATAC (Arimidex, Tamoxifen, Alone or in Combination) trial is a randomised, double-blind trial comparing 'Arimidex' (anastrozole), alone or in combination with tamoxifen, relative to tamoxifen alone as a five year adjuvant treatment for postmenopausal women with early breast cancer. Because tamoxifen is associated with endometrial pathology, the ATAC endometrial subprotocol was initiated to establish the background prevalence of pathology, and to assess prospectively the incidence and nature of intrauterine changes before and following endocrine therapy. Setting: International. Population and Study Design: Two hundred and eighty-five women entered the subprotocol: the mean age was 60 years (range 44-80 years); 113 women (40%) had taken hormone replacement therapy prior to randomisation, and 238 women were parous (84%). The age at onset of the menopause was 32-58 years, with the majority becoming menopausal between 46 and 55 years of age. Two hundred and seventy-two women had a hysteroscopy before they commenced trial medication. Hysteroscopy was performed successfully in 265 women. In six women, failure of hysteroscopy at baseline led to withdrawal from the study. Three of the women who withdrew had a pipelle biopsy taken. Therefore, the total number of endometrial biopsies at baseline was 268. Main outcome measures: To assess the demographic characteristics of women entering the endometrial subprotocol and their hysteroscopic and histological findings before commencing trial medication. Results: At hysteroscopy, there was a diagnosis of endometrial polyps in 34 women (13%), fibroids in 16 women (6%) and one case of suspicious endometrium, which was confirmed as a polyp on histology. Only 21 of the 34 polyps seen hysteroscopically were proven histologically (62% accuracy of hysteroscopy). Final histology found the prevalence of endometrial diagnostic categories as follows: 123 inactive endometrium (46%), 20 benign polyps (7%), 17 secretory endometrium (6%), 7 proliferative endometrium (3%), 3 atypical hyperplasia (2 in a polyp), 1 simple hyperplasia (in a polyp) and 1 fibroid. The remaining women had pipelle samples with insufficient tissue obtained, indicating a normal endometrial cavity. Conclusion: This is the first study of such size in gynaecologically asymptomatic breast cancer patients. This paper describes the findings in individual patients before any trial treatment was given. In this baseline group, 82% (219/268) of women had a normal endometrial cavity; 18% (49/268) had endometrial activity (proliferative or secretory endometrium in 9%) or an intracavity abnormality (hyperplasia, polyps and a fibroid in 9%). In total, 36% of biopsies had insufficient tissue for diagnosis, which in combination with a normal hysteroscopy was classed as normal. The appearance of a polyp hysteroscopically in this group was not proven histologically in approximately 40% of cases. The development of uterine pathology over time in the ATAC study will subsequently be assessed against the findings of this baseline paper.