We have at our disposal observations relating to the mortality due to different types of cancer and we wish to compare them with a theoretical model. The usual models lead to increased mortality with increasing age, instead of the drop which is observed after eighty. There are two reasons for the drop in mortality at high age. The first one is called "competing risk"; the older subjects can die from many other causes, even if they have the cancer under study. The second reason could be a protective effect in a proportion of the population.