This retrospective study was designed to establish some guidelines for the treatment of the neck in the case of supraglottic cancer. The patient population included a series of 264 patients, from 39 to 76 years old (mean age 58.3), who had undergone monolateral (76) or bilateral (188) neck dissection for a total of 452 neck dissections. The distribution of the patients, according to pT category, was the following: 17 T1 (6.4%), 88 T2 (33.3%), 107 T3 (40.5%) and 52 T4 (19.7%). In 121 patients the tumor was central (45.8%), whereas in 143 (54.2%) the neoplasm did not extend beyond the midline and was therefore defined as lateralized. The overall incidence of lymph node metastases was 43.4% (39 N1, 32 N2b, 28 N2c), and the rate of occult metastases was 24.7%. The distribution of metastases according to pT category was as follows: 6.2% T1, 30.7% T2, 38.3% T3 and 57.7% T4 (p < 0.001). Occult metastases distribution was: 0% T1, 19.6% T2, 26.7% T3 and 44.4% T4 (0.001 < p < 0.01). The incidence of bilateral metastases was significantly different (p < 0.001) in central (45.8%) and lateralized tumors (7.8%). The distribution of metastases according to level was 0.8% I, 82.4% II, 35.2% III, 13.6% IV and 0% V (p < 0.001). When level I or IV was involved, lymph node metastases were also present in level II and/or III. These results suggest that contralateral elective neck dissection is not required in lateralized tumors and elective neck dissection is not indicated in T1 lesions. Since no occult metastases were detected in level I or V, the management of choice for the clinically negative neck might well be a selective dissection limited to levels II, III and IV ("lateral neck dissection"). Levels I and V should be dissected only when metastatic nodes are found.