OBJECTIVES To examine national variation in use of the Medicare hospice benefit by older individuals before their death, and to identify individual characteristics and local market factors associated with hospice use. DESIGN Retrospective analysis of Medicare administrative data. SETTING Hospice care. PARTICIPANTS Older Medicare enrollees who died in 1996. MEASUREMENTS Rate of hospice use per 1,000 older Medicare beneficiary deaths. RESULTS Overall, 155 of every 1,000 older Medicare beneficiaries who die use hospice before death. This rate is significantly higher among younger older persons (P < .001), non-blacks (P < .001), persons living in wealthier areas (P < .001), and persons in urban areas (P < .001). Areas with a higher proportion of non-cancer diagnoses among hospice users have higher rates of hospice use for both cancer and non-cancer reasons than areas with a majority of hospice users having cancer diagnoses (P < .001). Hospice use is higher in areas with fewer hospital beds per capita (P < .001), areas with lower in-hospital death rates (P < .001), and areas with higher HMO enrollment (P < .001). Rates of hospice use are also positively related to average reimbursements for health care (P < .001) and to physicians per capita (P < .001). In the largest metropolitan statistical areas (MSAs), rates of hospice use vary more than 11-fold from a low of 35.15 (Portland, ME) to a high of 397.2 per 1,000 deaths (Ft. Lauderdale, FL). CONCLUSIONS The wide variation in hospice use suggests that there is great potential to increase the number of users of the Medicare hospice benefit.