In nineteenth-century Canada, a Census was conducted every ten years, beginning in 1851. Historians have tried to use the information in the local, provincial, and national Censuses to determine the rate of mortality and causes of death. The mortality statistics are thought to be reliable after 1921, but practical as well as conceptual problems interfere with attempts to establish retrospective mortality patterns using the nineteenth-century Census. In this paper, I will review the problems with this source on mortality. I will then compare the Census data with that derived from a study of a physician's daybooks, focusing on four specific areas of discrepancy and the possible reasons for the differences. Finally, I will suggest that the combined use of both types of sources may challenge some durable assumptions and could help to establish more reliable estimates of mortality patterns.