Unregistered temporary migration can create large discrepancies between death rates according to the legal, factual or habitual residence of the deceased. The most accurate death rates are those that refer to population numbers based on habitual residence, as they concern the population fully exposed to the hazards of a specific municipality. Using the Belgian population census, the register of vital events and individual death certificates, we calculate the death rates of the population with habitual residence in three large Belgian cities (Brussels, Liege and Schaarbeek) around 1910 and compare them with the typically calculated rates based on de facto deaths. Most significantly, the presence of medical institutions in large cities such as Brussels and Liege artificially increased the rates based on deaths within their territory. In the suburb of Schaarbeek, on the other hand, large numbers of people with habitual residence died outside the town. Consequently, our results suggest that for some age groups, Brussels' well-known excess mortality during the nineteenth and early twentieth centuries can partly be attributed to the large number of people without habitual residence who died in the capital.