Aim: To analyse causes of failure of sentinel node (SN) procedures in breast cancer patients and assess the role of pre-operative ultrasound examination of the axilla. Methods: In 138 consecutive clinically node negative breast cancer patients with the primary tumour in situ a SN procedure with radiolabeled colloid and blue dye was performed. Radioactivity in the SN was scored as inadequate or adequate. The axillary lymph node dissection scored for number of involved nodes and presence of extranodal growth. Results: In 53/138 patients, the SN was positive for tumour. Full axillary node dissection revealed that 58/138 were node positive. So in five patients the SN failed to predict true nodal status. In 3/5, the radioactive ratio (SN vs background) was inadequate. All were found to have extensive nodal involvement. The radioactivity ratio was inadequate in 37/138 patients. This ratio was inadequate in 10 of 15 patients with ≥4 positive nodes and 27 of 123 in patients with 0-3 positive nodes (p<0.001). If extranodal growth was present the radioactive ratio was inadequate in 13 of 18 patients, whilst this was only the case in 24 of 120 patients without extranodal growth or metastases (p<0.001). Ultrasound (US) examination and US-guided FNAC was able to pre-operatively identify 16 of the 26 patients with four or more metastases in the axilla. Conclusions: Extensive nodal involvement is an important cause of failure of the sentinel node biopsy. Pre-operative ultrasound examination of the axilla can avoid this in almost two thirds of these patients.