The aim of the study was to evaluate a hospital-based injury recording system on hip fracture registration in elderly persons aged + 65 years from 1994 through 2008, and to examine the agreement between the number of validated fractures and the number of fractures reported to the Norwegian Patient Registry using three different sources: (1) Medical records, (2) Patient administrative system and (3) The hospital's hip fracture record to the Norwegian Patient Registry from 2002 through 2008. The injury recording system included 582 hip fracture events and 535 (92%) were confirmed through the medical records. Reasons for non-verification were different coding failures. Searching the patient administrative system using ICD codes identified 16 hip fractures not included in the fracture registry between 2002 through 2008. The total number was the same as the number of hip fractures reported to the Norwegian Patient Registry using ICD codes alone for identification. The conclusion is that on well-defined diagnosis like hip fractures, local fracture registries may obtain a high degree of reliability if different sources are available for quality control. Well-functioning patient administrative systems may be used to study numbers of hip fractures.