OBJECTIVE: To evaluate the diagnostic accuracy of fluorodeoxyglucose positron emission tomography (FDG-PET) relative to computed tomography (CT) for detecting metastatic cervical lymph nodes in patients with squamous cell carcinoma of the head and neck (HNSCC), and to ascertain the factors that affect this accuracy. METHODS: A total of 1076 lymph nodes obtained from 35 neck dissections in 26 HNSCC patients who preoperatively underwent both FDG-PET and CT were retrospectively analyzed. For pathological metastatic lymph nodes, the lymph node size (short-axis diameter), the ratio of intranodal tumor deposits, and the size of intranodal tumor deposits (maximum diameter of metastatic foci in each lymph node) were histologically recorded. RESULTS: Forty-six lymph nodes from 23 neck sides were pathologically diagnosed metastases. The sensitivity, specificity, accuracy, positive predictive value, and negative predictive value of FDG-PET evaluated individually per neck side were 74%, 92%, 80%, 94%, and 65%, respectively, whereas those of CT were 78%, 58%, 71%, 78%, and 58%, respectively. FDG-PET detected 100% of metastatic lymph nodes > or =10 mm, intranodal tumor deposits > or =9 mm, and intranodal tumor deposits with a ratio >75%, whereas no nodes or tumor deposits smaller than 5 mm were detected. The spatial resolution limitations of FDG-PET were responsible for 16 of 20 (80%) false-negative PET results in lymph nodes. CONCLUSIONS: FDG-PET is a useful tool for preoperative evaluation of the neck because it accurately detects metastatic lymph nodes > or =10 mm and has fewer false-positive cases than CT. The high specificity of FDG-PET for lymph node metastases may play an important role in avoiding unnecessary neck dissection.