Large randomized clinical trials have redefined patterns of spread which have major implications for treatment options and outcomes in early stage breast cancer. Trials comparing mastectomy plus radiotherapy with mastectomy alone in women with T1-2 N0-1 M0 disease show clearly that the probability of developing distant metastases is unaffected by treatment of the lymphatic pathways. The inference is that, for the vast majority of women presenting with early stage disease, the cancer is either confined to the breast or else disseminated via vascular and lymphatic channels more or less simultaneously. Breast cancer is almost unique in that distant metastases may remain occult for several decades before becoming clinically apparent by mechanisms which remain unclear. The probability of haematogenous dissemination correlates closely with tumour size and this observation is the basis of population-based screening programmes discussed elsewhere in this issue. By identifying tumours earlier in their evolution, effective treatment of the primary disease is expected to be curative in a greater proportion of women. The implications for breast conserving management of women with early stage breast cancer are that initial surgery and radiotherapy must be of a high quality and promptly delivered if the expected reductions in breast cancer mortality are to materialize.