INTRODUCTION: Falling when older is a major public health issue. There is compelling evidence to show that specific exercise programs can reduce the risk and rate of falls in community-dwelling older people. Another major health issue for older people living in the community is upper limb dysfunction, including shoulder pain. Home-based exercise programs appeal to some older people, due to their convenience. RESEARCH QUESTIONS: This trial aims to determine the effectiveness and cost-effectiveness of a home-based lower limb exercise program compared with a home-based upper limb exercise program to prevent falls and upper limb dysfunction among community-dwelling people aged 65+ years. DESIGN: Randomised, controlled trial. PARTICIPANTS AND SETTING: A total of 576 community-dwelling people will be recruited from the Illawarra and Shoalhaven regions of New South Wales, Australia. INTERVENTION: Participants will be randomised to either a home-based lower limb exercise intervention or a home-based upper limb exercise intervention. The lower limb program is designed to improve balance and strength in the lower limbs. The upper limb program is designed to improve upper limb strength and mobility. Participants will attend three group-based instruction sessions to learn and progress the exercises, and will be instructed to perform the exercises three times per week at home for 12 months. OUTCOME MEASURES: The two primary outcomes will be fall rates, recorded with monthly calendars for a 12-month period, and upper limb dysfunction, measured with the Disability of the Arm, Shoulder and Hand questionnaire. Secondary outcomes will include: lower limb strength and balance; shoulder strength and mobility; physical activity; quality of life; attitudes to exercise; proportion of fallers; fear of falling; and health and community service use. The cost-effectiveness of both exercise programs from a health and community service provider perspective will be evaluated. ANALYSIS: Negative binomial regression models will be used to estimate the between-group difference in fall rates. Modified Poisson regression models will be used to compare groups on dichotomous outcome measures. Linear regression models will be used to assess the effect of group allocation on the continuously scored measures, after adjusting for baseline scores. Two economic evaluations will be conducted: the first will assess cost-effectiveness of the lower limb program compared with the upper limb program; and the second will assess cost-effectiveness of the upper limb program compared with the lower limb program. DISCUSSION: If effective, the trial will provide a model for both upper limb and lower limb exercise programs that can be performed at home and implemented at scale to community-dwelling older adults.