A high excess mortality is well described after hip fracture. Deaths are in part related to comorbidity and in part due directly or indirectly to the hip fracture event itself (causally related deaths). The aim of this study was to examine the quantum and pattern of mortality following hip fracture. We studied 160,000 hip fractures in men and women aged 50 years or more, in 28.8 million person-years from the patient register of Sweden, using Poisson models applied to hip fracture patients and the general population. At all ages the risk of death was markedly increased compared with population values immediately after the event. Mortality subsequently decreased over a period of 6 months, but thereafter remained higher than that of the general population. The latter function was assumed to account for deaths related to comorbidity and the residuum assumed to be due to the hip fracture. Causally related deaths comprised 17-32% of all deaths associated with hip fracture (depending on age) and accounted for more than 1.5% of all deaths in the population aged 50 years or more. Hip fracture was a more common cause for mortality than pancreatic or stomach cancer. Thus, interventions that decreased hip fracture rate by, say, 50% would avoid 0.75% or more of all deaths.