Axillary lymph node involvement is the best prognostic factor for breast cancer survival. Staging breast cancers by axillary dissection remains standard management and is part of the UK national guidelines for breast cancer treatment. In the presence of involved axillary lymph nodes best treatment has been shown to be axillary clearance (Fentiman and Mansell, 1991), but clearly for women whose nodes are uninvolved avoidance of morbidity is optimal and this will be achieved by minimal dissection of the axilla. Thus, for node-negative women the introduction of the sentinel node biopsy technique may revolutionise the approach to the axilla. These will be women with mammographic screen detected small well and moderately differentiated tumours (Hadjiloucas and Bundred, 2000). The impact of sentinel node biopsy in women who have symptomatic large tumours is unproven, and around half of these women will require a second procedure to clear their axilla or radiotherapy as treatment. Even for those women found to have involved sentinel lymph nodes the ability to use early systemic chemotherapy followed by axillary clearance or radiotherapy may provide long-term survival gains. Sentinel node biopsy should not, however, become routine practice until randomised controlled trials have proven its benefit and safety in reducing morbidity. Several randomised controlled trials (including ALMANAC) are currently underway. British Journal of Cancer (2002) Sentinel node lymphoscintigraphy leads to the visualisation of hot radioactive nodes in sites other than the axilla in about 13% of all cases (range 2 -35% in series using different injection techniques) (Cserni and Szekeres, 2001). Tanis and co-workers in this issue of the Journal report the detection of non-axillary sentinel nodes and its impact on treatment. Earlier studies where internal mammary (IM) lymph node dissection has been performed have shown IM node involvement in about 23% of breast cancer patients (Cserni and Szekeres, 2001). In the study by Tanis et al (this issue) the proportion of detected non-axillary node metastases was about 5% among all patients, which is similar to findings in other studies where lymphoscintigraphy has been used (Cserni and Szekeres, 2001;Dupont et al, 2001). Only patients with a visualised non-axillary sentinel node were biopsied but this low proportion of metastases might also reflect that breast cancer is detected at an earlier stage today. The proportion of involved nodes was 20 -25% among those patients who underwent an IM or other nonaxillary sentinel node biopsy.Centrally and medially located tumours have been reported to have a higher proportion of IM lymph node metastases compared to lateral tumours (Manji, 1982). However, in the overview of series undergoing extended radical mastectomy (Cserni and Szekeres, 2001) the range of IM lymph node involvement was 13.3 -35.3% in lateral tumours and 19.5 -32.6% in central and medial tumours. Tumour location alone is therefore not a reliable indicator of risk for IM lymph node metastases (Ur...