Surgery is one of the mainstays of head and neck cancer treatment, and aims at radical resection of the tumor with 1 cm tumor-free margins to obtain locoregional control. Surgical margins are evaluated by histopathological examination of the resection specimen. It has been long an enigma that approximately 10–30% of surgically treated head and neck cancer patients develop locoregional recurrences even though the resection margins were microscopically tumor-free. However, the origins of these recurrences have been elucidated by a variety of molecular studies. Recurrences arise either from minimal residual disease, cancer cells in the surgical margins that escape detection by the pathologist when examining the specimen, or from precancerous mucosal changes that may remain unnoticed. Head and neck tumors develop in mucosal precursor changes that are sometimes visible but mostly not, fueling research into imaging modalities such as autofluorescence, to improve visualization. Mostly unnoticed, these precancerous changes may stay behind when the tumor is resected, and subsequent malignant progression will cause a local relapse. This led to a clinical trial of autofluorescence-guided surgery, of which the results were reported in 2020. This review focuses on the most recent literature of the improved diagnosis of the resection margins of surgically treated head and neck cancer patients, the pathobiological origin of recurrent disease, and relevant biomarkers to predict local relapse. Directions for further research will be discussed, including potential options for improved and personalized treatment, based on the most recently published data.