The Registrar-General’s Returns are an integral source for historical demographers. Concerns have been raised however, over the geographical accuracy of their pre-1911 mortality figures when institutional deaths were not redistributed to place of residence. This paper determines the extent of the distortions caused by institutional mortality in the context of aggregate infant mortality rates for London’s registration sub-districts. The potential of two alternative methods to ‘correct’ these distortions is then assessed. The first method uses indirect estimation techniques based on data from the 1911 Fertility Census, and the second exploits the rich detail available from the Medical Officer of Health Reports. Through narrowing the focus to seven London registration sub-districts over the years 1896-1911, it is shown that both suggested alternative methods remove the institutional mortality biases found in the Registrar-General’s figures, yet they come with their own limitations.