Data from population-based cancer registries provide information on the causes and outcome of cancer and form a basis for important decisionmaking in connection with the prevention of cancer and the planning of health services. This makes it of the utmost importance to assess the data at all stages of collection to ensure the highest possible quality. The present study focuses on the quality of the Cancer Registry of Norway's data on head and neck cancer for the period 1953–1991. When the study was started, 16,104 cases of head and neck malignancies had been registered. All histological codes were reviewed. The pathologists' reports were reevaluated for 369 cases selected according to set criteria: 133 cases received a new histological code without being exluded from the data material; 112 cases were excluded. The distribution of histological diagnoses for each location is presented. A reevaluation of 300 cases selected at random from the corrected series indicates discrepancies between the pathologist's classification and the Registry's coding in less than 2% (1.4%) of all cases. The percentage that lacked histological verification fell from 5.7% in the first decade to 2.1% during the last 9-year period. Completeness of the Cancer Registry's data base was checked against hospital-based registries and this investigation showed that virtually all new cases are reported. We conclude that the data on head and neck cancer for the studied time period meet standards that justify their use as a basis for epidemiological as well as clinical studies.