INTRODUCTION: The lack of registration of women who have received no or alternative treatment for breast cancer has been criticised. No distinction is made in the Danish Cancer Register between these patients and those who only receive palliative treatment for other reasons, such as old age, advanced disease, and competing illnesses. We have estimated the number of women in this group of patients, who, in reality, had not received any treatment with the intent to cure under the health care system, and whether a meaningful analysis of survival for these patients is feasible. METHOD: All women with breast cancer diagnosed during the years 1978-1995 were extracted from the Cancer Register, and we isolated those who had been registered as having had no or only palliative treatment and who had survived for a minimum of 45 days after diagnosis. A search was made in the Danish Breast Cancer Co-operative Group register for unreported treatment and the residual group was followed up individually. RESULTS: Out of 49.058 women with histologically or cytologically verified breast cancer, the Cancer Register listed 840 women with no registered treatment of their disease. Of these, there were 103 cases of carcinoma in situ. A match with the DBCG register revealed that 188 women had nevertheless been operated on. Among the remaining 549 women, 99 were truly untreated, and for 77 of these the reasons given were another or advanced disease or old age. Only 22 women had initially declined treatment for no specific reason. Five of these had later decided on subsequent curative treatment, which leaves 17 women in the category "breast cancer untreated at her own request" ("untreated breast cancer at own will"). Nine are dead, five had their tumour excised at biopsy, and the remaining three are alive with tumours diagnosed by fine needle aspiration biopsy (1) or thru-cut biopsy (2) after 7.7 and 4 years, respectively. CONCLUSION: This report has shown that a survival analysis based on the Cancer Register of untreated breast cancer in relation to treated breast cancer is not meaningful. A true estimation of survival after untreated versus treated breast cancer can only be achieved through a randomised study, which would be unethical.