Patients with limited distant metastatic disease, also known as oligometastasis, show better survival rates than polymetastatic patients, and may be amenable for curative-intent treatment. The definition of oligometastasis, however, is unknown, and no quantitative analyses on the cutoff value for oligometastasis have been performed before. This study aims to derive specific threshold values for the number of metastases and affected locations that defines oligometastatic disease in head and neck squamous cell carcinoma. A retrospective cohort study was conducted including all patients diagnosed with distant metastases between 2006 and 2021. For each patient, the number of distant metastases and affected locations was recorded on the basis of the available imaging at the time of diagnosis. Cox regression analyses and a machine-learning k-means algorithm were used to determine threshold values. A total of 384 patients untreated for their metastatic foci were analyzed. Most patients (n=207; 53.9%) had metastasis to one anatomic location, followed by metastases in two anatomic locations (n=62; 16.1%). The majority of patients had ≥9 metastatic foci (n=174; 45.3%), followed by one focus (n=74; 19.3%) and two foci (n=32; 8.3%). Cox regression and machine-learning k-means models showed that although the number of metastases did not predict survival, the number of affected locations did significantly (p<.001), by identifying a threshold of two locations. Contrary to the prevalent dogma, the definition of oligometastasis should not be defined by the number of metastases but rather by the number of affected locations, with a maximum number of affected locations set at two.