Incidence, prevalence and mortality are commonly used measures to assess the impact of disease on human populations. Prevalence, although regularly assessed for a number of different diseases, has only had recent use to measure the impact of cancer. The calculation of the prevalence of cancer presents several difficulties since there is no reporting mechanism established to measure the proportion of the community that has the disease. In the absence of such a mechanism, mortality data linked to incidence data from cancer registries have been used. The assumption is made that once diagnosed with cancer an individual remains a prevalent case until death. In this paper we present alternatives to this assumption and use them to produce estimates of cancer prevalence. We illustrate the effect of these assumptions on the calculated prevalence of cancer using data from the British Columbia Cancer Registry.