Most studies concerning the use of the sentinel node technique in head and neck cancers have included clinically N0 patients with primary early stage tumours of the oral cavity or upper part of oropharynx; furthermore, node sampling has been performed during the same session, but separately from the tumour. The perspective of avoiding unnecessary neck dissection, without increasing the risk of delayed diagnosis of lymph node metastasis, is rewarding, not only for early stage tumours of the oral cavity but also for tumours in advanced stages and/or at different anatomic sites. In the attempt to establish the reliability of extended use of the sentinel node technique, 100 consecutive untreated patients (from 1999 to 2002) with tumours located in the oral cavity, oropharynx, hypopharynx and larynx, at any T stage, entered the study. N+ patients with paramedian tumours and contralateral clinically negative nodes were also enrolled. After injection of the 99mTc albumin microcolloid, pre- and intra-operative evaluations with a gamma-probe were done. N0 patients (59) were submitted to mono- or bilateral selective neck dissection; the N+ patients (41) received homolateral dissection of all levels and contralateral selective dissection. An en bloc resection of the tumour was performed both in N0 and N+ patients. In the N0 group, histological examination showed no evidence of metastases in "hot" nodes in 34 patients and also the remaining nodes were negative. Metastases were found in one or more of the gamma-probe positive nodes (14 cases), or in a closely located node at the same level (2 cases) or in a node close to a "hot" area of the submandibular salivary gland (1 case). In 8 patients, lymphoscintigraphy did not identify any sentinel node and histology of all lymph nodes was negative for metastases. In the N+ group, no metastases were found in the sentinel nodes of 21 patients and also the remaining nodes were negative; in 4 patients, metastases were found in sentinel nodes. In 16 patients, lymphoscintigraphy did not identify any sentinel node and histology of all lymph nodes was negative for metastases. In no patients were metastases found outside the level containing the lymph node identified as sentinel by the gamma-probe. In conclusion, the strategy of the sentinel node is reliable, but, to be confirmed as a standard approach, it requires trials with a larger number of patients. The technique requires a multidisciplinary and well "amalgamated" team. It may likely be used also in T3 and T4 oro-hypopharyngeal and laryngeal primary tumours and to determine surgical treatment of the contralateral neck in patients with N2a, N2b, N3 on T close to the midline.