Cancer ranks among the most formidable diseases. Currently, the treatment of malignant tumors has entered the immunotherapy era. Immunotherapy has achieved remarkable progress in treating malignant tumors, including head neck squamous cell carcinoma. Nevertheless, a significant number of patients exhibit a limited response to this treatment. Thus, the quest for novel molecular biomarkers to assess the efficacy of immunotherapy is of utmost importance. In recent years, the prediction and evaluation of immune efficacy have emerged as focal points of research. Biomarkers developed based on tissue biopsies (such as programmed death ligand-1 expression, tumor infiltrates lymphocyte subsets, tumor mutation burden, cancer-associated fibroblasts, etc.), liquid biopsies (circulating tumor DNA, circulating tumor cells, and extracellular vesicles, etc.), when combined with nanotechnology, have shown the potential for highly sensitive prediction. This is achieved through non-invasive real-time monitoring of clonal evolution and immune escape. Moreover, radiomics and artificial intelligence (such as deep-learning models) can noninvasively predict and evaluate treatment response and prognosis. In this study, we comprehensively summarize the research progress of molecular markers for predicting and evaluating the efficacy of immunotherapy in head neck squamous cell carcinoma. We approach this from the perspectives of tissue biopsy, liquid biopsy, radiomics, and artificial intelligence.