Epidemiological cancer data shed light on key questions within basic science, clinical medicine and public health. For decades, Denmark has had linkable health registers that contain individual level data on the entire population with virtually complete follow-up. This has enabled high quality studies of cancer epidemiology and minimized the challenges often faced in many countries, such as uncertain identification of the study base, age misreporting, and low validity of the cancer diagnoses. However, methodological challenges still remain to be addressed, especially in cancer epidemiology studies among the elderly and the oldest-old. For example, a characteristic pattern for many cancer types is that the incidence increases up to a maximum at about ages 75-90 years and is then followed by a decline or a leveling off at the oldest ages. It has been suggested that the oldest individuals may be asymptomatic, or even insusceptible to cancer. An alternative interpretation is that this pattern is an artifact due to lower diagnostic intensity among the elderly and oldest-old caused by higher levels of co-morbidities in this age group. Currently, the available cancer epidemiology data are not able to provide clear evidence for any of these hypotheses.