This study was aimed at investigating the role of CT in the surgical planning (partial vs. total laryngectomy) of supraglottic laryngeal carcinoma. Sixty-nine patients affected with supraglottic cancers were reviewed to assess the accuracy and the clinical role of CT and laryngoscopy. The detection rates of tumor spread to the glottis, to thyroid and arytenoid cartilages, pyriform sinus and base of the tongue were compared and correlated with surgical and pathologic data. Thirty-one patients underwent supraglottic laryngectomy and 38 had total laryngectomy. Endoscopy correctly assessed the glottis in 54/69 patients (78.2%) and CT in 52/69 (75.3%). The two techniques were in agreement in 47/69 patients: 42/47 (89.4%) had a correct diagnosis. Sensitivity, specificity, accuracy, positive and negative predictive values in assessing neoplastic invasion at the glottic level were, respectively, 68.4%, 90.3%, 78.2%, 89.6%, 94.1% and 86.6% when endoscopy and CT were in agreement. In deciding the feasibility of supraglottic laryngectomy, the results of endoscopy alone did not differ significantly from those of CT and endoscopy in agreement (X2 = 3.255, p > 0.05), whereas the negative predictive value of CT was significantly lower than that of the two techniques in agreement (X2 = 4.55, 0.05 > p > 0.025). In our experience, CT did not significantly change the surgical treatment planned on the basis of endoscopic findings. Therefore, when assessing the feasibility of partial laryngectomy for supraglottic carcinoma, CT cannot be considered a cost-effective tool. Nine of 33 (27.2%) patients treated with total laryngectomy had local recurrences at the hypopharynx, probably because the primary tumor was underestimated at both preoperative staging and during surgery.