Non-invasive breast cancer can be of either the ductal or the lobular type. While the former is often associated with progression to invasive cancer at the same site, lobular carcinoma in situ (LCIS) is a risk factor for invasive cancer rather than a precursor. As a result, LCIS has been used as one entry criterion for the International Breast Intervention Study in which women at increased risk because of histological findings or family history are randomised to receive either tamoxifen 20 mg daily or placebo for 5 years. Four randomised trials have examined treatment options for ductal carcinoma in situ (DCIS) and all have demanded complete local excision as a necessary qualification for entry. Because of this, up to half the cases of DCIS were ineligible for entry since the disease was too extensive and was therefore usually treated by total mastectomy. In most studies--NSABP B-17, EORTC 10853 and Swedish trial--the randomisation was between breast irradiation and observation. Only one trial, NSABP B-17, has published results, and these suggest that progression to invasive cancer is reduced by irradiation. Withholding radiotherapy did not affect mortality. The UK DCIS trial is comparing the effects of both radiotherapy and tamoxifen, but as yet no results are available. At present, radiotherapy should not be used as standard treatment of completely excised DCIS. Appropriate indications for irradiation will be determined when results of histologically characterised cases participating in mature trials become available. Total mastectomy remains the standard treatment for extensive DCIS, but the next generation of trials may examine the role of endocrine manipulation in cases with estrogen-receptor-positive tumours.