BACKGROUND: Older people living with dementia prefer to stay at home to receive support. But they are at high risk of death and/or hospital admissions. This study primarily aimed to determine risk factors for time to death or hospital admission (combined) in a sample of community-dwelling older people living with dementia in Australia. As a secondary study purpose, risk factors for time to death were also examined. METHODS: This study used the data of a previous project which had been implemented during September 2007 and February 2009. The original project had recruited 354 eligible clients (aged 70 and over, and living with dementia) for Extended Aged Care At home Dementia program services during September 2007 and 2008. Client information and carer stress had been collected from their case managers through a baseline survey and three-monthly follow-up surveys (up to four in total). The principal data collection tools included Global Deterioration Scale, Modified Barthel Index, Instrumental-Dependency OARS, Adapted Cohen-Mansfield Agitation Inventory, as well as measures of clients' socio-demographic characteristics, service use and diseases diagnoses. The sample of our study included 284 clients with at least one follow-up survey. The outcome variable was death or hospital admission, and death during six, nine and 16-month study periods. Stepwise backwards multivariate Cox proportional hazards analysis was employed, and Kaplan-Meier survival analysis using censored data was displayed. RESULTS: Having previous hospital admissions was a consistent risk factor for time to death or hospital admission (six-month: HR = 3.12; nine-month: HR = 2.80; 16-month: HR = 2.93) and for time to death (six-month: HR = 2.27; 16-month: HR = 2.12) over time. Previously worse cognitive status was a consistent risk factor over time (six- and nine-month: HR = 0.58; 16-month: HR = 0.65), but no previous use of community care was only a short-term risk factor (six-month: HR = 0.42) for time to death or hospital admission. CONCLUSIONS: Previous hospital admissions and previously worse cognitive status are target intervention areas for reducing dementia clients' risk of time to death or hospital admission, and/or death. Having previous use of community care as a short-term protective factor for dementia clients' time to death or hospital admission is noteworthy.