al., 1988) comparing surgery with tamoxifen did not find a difference in survival between the two groups but locoregional control led to a better result in the surgically treated patients. This led to the conclusion that optimum treatment for elderly patients could include both surgery and tamoxifen. A randomised study (Bates et al., 1991) compared tamoxifen only with optimal surgery followed by adjuvant tamoxifen. The overall survival after 3 years was similar between the two groups but a statistically significant higher locoregional relapse rate was reported in the group treated by tamoxifen only.The rationale of the present study was that non-randomised clinical studies (Helleberg et al., 1982;Preece et al., 1982;Allan et al., 1985) showed that tamoxifen as first-line treatment was of value in a selected group of elderly breast cancer patients. The question was whether surgery without axillary dissection combined with tamoxifen in patients with breast cancer and without clinical nodal involvement could give the same results in terms of overall survival, disease-free survival and quality of life compared with more radical treatments.In a previous paper (Martelli et al., 1993) (Hermanek et al., 1987), 77 (24%) a T2, 1 (0.3%) a T3 and 24 patients (7.5%) presented with a tumour infiltrating the skin but not the underlying muscle (T4b). All patients had surgery performed under local anaesthesia without axillary dissection; 298 underwent wide lumpectomy or quadrantectomy and 23 total mastectomy. The conservative surgical techniques have been previously described (Veronesi et al., 1990;Galante et al., 1992).All patients underwent resection of the tumour with removal of at least 2 cm of normal tissue to ensure a specimen with tumour-free margins. Patients with the margins of resection in tumour tissue were excluded from the analysis, since they were candidates for a re-excision or radiotherapy.Independent of hormone receptor status, all patients received indefinitely 20 mg tamoxifen daily from the time of surgery. Tumour specimens were assayed for both oestrogen receptor (ER) and progesterone receptor (PgR) levels by using the dextran-coated charcoal Scatchard analysis. RecepCorrespondence: G Martelli