OBJECTIVES: Patients with H&N cancer carry the highest risk of secondary primary cancers. Determining the underlying cause of death is conflicting in presence of multiple primary cancer sites, and the actual burden of H&N cancers may be underestimated by causes of death statistics. METHODS: Using the French National Hospital Discharge (PMSI) database, we identified all adult patients residing in Metropolitan France and diagnosed with H&N cancer (ICD-10: C00-C06; C09-C14; C30.0; C31; C32) in 2008-2012. Among patients who died at hospital, we considered advanced H&N cancer (stage III/IV at diagnosis or disease progression) as a cause of death and recorded competing causes of death from other primary cancer sites and Charlson comorbidities. Death records from PMSI were compared to National causes of death statistics (CEPIDC) using the same ICD-10 definitions. The mortality gap (PMSI minus CEPIDC) was described overall and by sex, age category, region, H&N cancer site, and year. The role of other primary cancer sites and Charlson comorbidities on misclassification bias (i.e., a positive mortality gap) was further explored in multivariate Poisson regression. RESULTS: Of 94,672 French patients identified with H&N cancer in 2008-2012, 41,503 patients with advanced H&N cancer died at hospital. In comparison, 25,647 (61.8%) deaths were attributed to H&N cancers in National causes of death statistics. Misclassification bias increased from 2008 to 2012 (+8% by year) and varied by H&N cancer site (lowest: lip; highest: oropharynx). In multivariate Poisson regression, misclassification bias significantly increased with records of another cancer before or at diagnosis of H&N cancer, distant metastasis, and palliative care. Correcting for mortality outside hospital in the PMSI discharge database would worsen the mortality gap. CONCLUSIONS: The study results suggest that the actual burden of H&N cancer is underestimated by more than 38% by National causes of death statistics. Misclassification bias increased in the recent years.