The study of long-term change has traditionally been severely hampered by changes in the administrative geography for which data have been published. The traditional response to this problem has been aggregation, however, this causes a loss of detail and may also lead to unreliable results due to the modifiable areal unit problem. This paper presents a methodology that allows longitudinal analysis to be performed with approximately the level of spatial detail for which the data were originally published. This is done using a Geographical Information System (GIS) to interpolate data published at different dates and for different reporting units onto a single, standardised administrative geography. Once this has been done, it becomes possible to compare many datasets over the long-term. Net migration is used as an example as this is calculated by comparing the populations of age and sex cohorts at two dates and the number of deaths in that cohort between the two dates. It is, therefore, calculated as a residual and is highly sensitive to population change caused by boundary changes. These will appear as errors in the net migration rate. The methodology relies on a large boundary GIS that has not yet been completed. For this reason the county of Gloucestershire in the south-west of England is used as an example. Once the GIS is completed, however, it will be possible to calculate spatially detailed net migration rates for the whole of England and Wales from the mid-nineteenth century to the present, sub-divided into ten-year age bands and by gender.