The value of elective lymph node dissection (ELND) for melanoma patients with clinically uninvolved regional nodes remains controversial. However, it has been proposed that selective 'sentinel' lymph node biopsy reliably identifies individuals with micrometastases, who are most likely to benefit from full ELND. The aim of this study was to confirm that metastatic melanoma cells travelling in lymphatics do not bypass the sentinel node. After preoperative lymphoscintigraphy and intraoperative injection of blue dye around the primary melanoma site, sentinel node biopsy was performed in 118 melanoma patients for whom full ELND was planned as part of their definitive surgical treatment. A confidently identified sentinel node was tumour positive in 22 out of 105 regional lymph node fields (21%). In 18 cases the sentinel node was the only node found to be involved and in four cases, additional nodes were positive. In two other patients a positive node was found when the sentinel lymph node had been negative. However, in each case an avoidable error of technique had occurred and definite blue staining indicated that the positive node was in fact another sentinel node. This study thus confirms that sentinel lymph node status reliably indicates whether metastatic melanoma is present in regional lymph nodes.