The proportion of the population over the age of retirement has risen in many countries, and this means there is a corresponding rise in the incidence of hip fractures. However, in order to reliably investigate the ability of interventions to prevent fracture, there needs to be a reliable measure of the incidence of hip fracture. The purpose of this study was to examine the inclusion and exclusion criteria used to identify hip fracture from hospital admission data and to examine the impact that these criteria have on estimated incidence of hip fracture. We examine the influence of: individual compared to consultant episode data; primary data compared to any diagnosis of hip fracture; emergency compared to elective admissions; and the influence of type and rate of surgery on incidence estimates. The results showed that classifying hip fractures by use of consultant episodes overestimated the rate of hip fracture by 6-31%, and this overestimation has increased in recent years. The use of primary diagnosis as opposed to any diagnosis underestimates hip fracture by 5%. Two percent (2%) of the people studied had an operation for a hip fracture but did not have a hip fracture diagnosis (many had a multiple fracture diagnosis), and 5.5% of the people studied had an elective admission for the hip fracture (perhaps falling in hospital during an elective admission). We conclude that the selection criteria can have a great influence on the number of hip fractures identified using routine data. There should be a standardized selection procedure for the identification of hip fracture, as this would enable interventions and preventive measures to be evaluated over time and facilitate comparisons of rates in different regions and countries, in order to examine factors associated with hip fracture.