The present study examined the Liverpool database in an attempt to determine what proportion of N0 necks for various head and neck primary sites harbored subclinical squamous cell carcinoma and whether empiric treatment of occult disease improved survival over and above a “wait-and-watch policy” (treatment when metastasis becomes manifest). One hundred seventeen neck dissections were carried out for N0 necks, with 32% of specimens found to contain squamous cell carcinoma. The risk of carcinoma was highest in the hypopharynx, with 50% of specimens associated with a pyriform fossa primary cancer. Twenty-nine percent of neck dissection specimens for oral cavity cancer contained carcinoma and this was commonly associated with lateral border of tongue or anterior floor of mouth carcinomas. Twenty-five percent of specimens when primary tumor was in the oropharynx contained carcinoma and were due to tonsillar carcinoma. Twenty-one percent of laryngeal cancers produced histologically positive nodes and were mostly associated with posterior epiglottic tumors. Two hundred forty-six patients had a pyriform fossa cancer and of these only 37 had N0 disease and surgical treatment. Of these, 23 patients had radical neck dissections, whereas in 14 the necks were not treated. There was no difference in survival between the two groups (ξ12 = 0.787, P = NS). The Liverpool database also contained 1631 previously untreated patients with no clinical evidence of neck node metastases. Of these only 107 had a neck dissection. There was no difference in survival (ξ12 = 2.79, P = NS). When these data were analyzed by multivariate methods (Cox's proportional hazards model) prophylactic neck dissection was found to have no significant effect.