A series of 478 patients with T,.,No glottic carcinoma treated by irradiation is presented. Of these patients, 320 were previously untreated, whereas 158 patients were referred for treatment of a recurrence after receiving radiotherapy elsewhere. The primary recurrence rate in the previously untreated patients was 10%. The rate was higher in T, and T, tumors, poorly differentiated tumors, and in patients who were in poor general condition. Over 80% of the recurrent tumors were Stage pT, or pT,, whereas 12% of total laryngectomy specimens showed necrosis only with no evidence of tumor. The necrosis rate in previously untreated patients was 1% for T, tumors, 4% for T, tumors, and 3% for T, tumors. Of all tumors, 60% were transglottic when they recurred, whereas only 29% were confined to the glottis at recurrence. Histologic diagnosis had a high sensitivity but a low specificity, indicating that a negative histologic report is unreliable. Of patients with a recurrent primary tumor, 13% were untreatable. The 5-year survival after a primary recurrence was 39%, and the main prognostic factors were sex, T stage at recurrence, and time to recurrence. Of patients available for follow-up at 5 years 49% were alive with a larynx, 5% were alive without a larynx, 13% were dead of the original cancer, and 33% had died of other causes. In those suffering a primary recurrence, the commonest cause of death was a subsequent lymph node metastasis, followed in order of frequency by stoma1 recurrence and recurrence in the pharyngeal remnant. The hospital mortality rate after laryngectomy was 3%, and 30% of patients undergoing laryngectomy developed a pharyngocutaneous fistula. The recurrence rate in lymph nodes was 14% at 5 years, general condition and T stage being the only significant predictors of recurrence. Only 17% of patients had small (N,) nodes by the time the diagnosis of cervical lymph node recurrence was made, and 27% of all patients were unsuitable for treatment. Host, tumor factors, and time to recurrence were not significant predictors of survival after node recurrence. The survival rate 5 years after node recurrence was IS%, and the main cause of death in those who died was uncontrolled disease in the neck. The hospital mortality after salvage neck dissection was 4.7%.