Objectives: To estimate the likelihood and costs of in-patient care in the last three years of life.

Methods: A population-based retrospective cohort study using linked hospital and death records to evaluate in-patient use by Western Australians who died in 2002.

Results: Age was unrelated to the likelihood of in-patient admission and inversely related to in-patient costs, after adjustment for sex, cause of death and proximity to death. In-patient costs increased in the final three quarters before death. In the last quarter before death, the predicted average quarterly in-patient cost increased 2.8 fold from quarter two and 3.8 fold from quarter three.

Conclusions: Older decedents were not more likely to be hospitalised than younger decedents in the final three years of life. Moreover, once hospitalised, their in-patient costs were lower. In-patient costs were heavily concentrated in the three last quarters of life.

Implications: Remaining lifetime is a significant predictor of in-patient costs. Failure to account for proximity to death will overemphasise the impact of population ageing on health care expenditure, because older people have a higher probability of dying.