Background The early and accurate detection of local or regional recurrence of head and neck tumor is critically important. Purpose To compare the diagnostic accuracy of contrast-enhanced computed tomography (CT), contrast-enhanced magnetic resonance imaging (MRI), and fluorodeoxyglucose (FDG) positron emission tomography (PET)-CT, alone and in combination, in detecting the locoregional recurrence of malignant head and neck tumor. Material and Methods A total of 93 patients with loco-regional recurrence of malignant head and neck tumors underwent CT, MRI, and PET-CT within 30 days before surgery. CT, MRI, and PET-CT for each patient were retrospectively reviewed to determine the presence of recurrent tumors in the primary site on a patient-by-patient basis and that of regional lymph nodes on a level-by-level basis. The diagnostic accuracy of CT, MRI, and PET-CT, alone and combined, were accessed with the postoperative histopathological findings or with 12-month follow-up results as the standard of reference. Results The sensitivity/specificity/and accuracy of CT, MRI, and PET-CT for the detection of primary site recurrence was 89.9/85.7/89.3%, 94.9/85.7/93.6%, and 97.5/92.9/96.8%, respectively. The sensitivity/specificity/accuracy of CT, MRI, and PET-CT for the detection of nodal recurrence was 66.3/99.4/92.4%, 74.7/99.4/94.2%, and 85.5/94.9/93.0%, respectively. MRI + PET-CT achieved the best performance in the receiver operating characteristics curve analysis (Az value = 0.958 for primary site recurrence and 0.929 for nodal recurrence). Conclusion MRI + PET-CT offered the highest diagnostic performance in the detection of loco-regional recurrence of malignant head and neck tumor, compared with CT, MRI, PET-CT, and other combinations including CT.