The decision regarding treatment of supraglottic carcinoma remains controversial, despite recent advances in radiotherapy and surgical procedures. The need to evaluate the prognostic importance of demographic, clinical, pathologic, and treatment-related variables prompted this study. The patient population was of 164 consecutive cases of supraglottic carcinoma. Clinical charts were reviewed and cases were restaged according to the 1987 version of the UICC-AJC classification based on the initial clinical description: 18 T1, 30 T2, 58 T3, 58 T4, 61 NO, 24 N1, and 79 N2a-N3. Management policy for these patients varied during the study, reflecting different opinions of the responsible physicians and technical advances. Seventy-seven patients (47%) underwent surgery, patients who underwent surgery and radiotherapy, respectively, remained alive with no evidence of disease. The 5-year actuarial survival rates were 58.5% for patients initially treated by surgery, and 16.3% for patients in the radiotherapy group. Although several demographic, clinical, and pathologic variables were studied, only initial treatment (p <.0001), N stage (p = .0003), and T stage (p = .0017) were deemed to have independent prognostic value by multivariate regression techniques based on Cox's proportional hazards model. This study has shown that survival of supraglottic carcinoma patients depends on the treatment modality, T stage, and N stage. Our results are disappointing with regard to radiotherapy as a treatment for supraglottic carcinoma in patients with tumors at clinical stages III and IV.