This study investigates the main features of collected disability statistics for the nationwide censuses in Sweden from 1860 to 1930. During this period, the disability prevalence rose from four to 21 individuals per thousand of the population. To understand this significant rise, there is a need to peer behind the census numbers to trace Swedish authorities’ approach to measuring disabilities in the population. We use qualitative methods to analyze the authorities’ means of collecting, categorizing and defining disability, while quantitative methods help us estimate the prevalence by disability type and gender across the study period in urban and rural areas. Our long-term findings reveal that the increase of disability prevalence in Swedish populations, and especially the peak in 1930, reflect a new approach to measuring disability. Authorities applied other methods to collect disability data by using additional sources, wider disability definitions, as well as introducing new categories. The temporal variations in disability prevalence were largely influenced by the censuses becoming more scientific in their reporting of disabilities, as well as by societal concerns and by what was viewed as disabling conditions when the censuses were taken.