This randomized controlled trial examined the effects of multicomponent training on muscle power output, muscle mass, and muscle tissue attenuation; the risk of falls; and functional outcomes in frail nonagenarians. Twenty-four elderly (91.9 ± 4.1 years old) were randomized into intervention or control group. The intervention group performed a twice-weekly, 12-week multicomponent exercise program composed of muscle power training (8-10 repetitions, 40-60 % of the one-repetition maximum) combined with balance and gait retraining. Strength and power tests were performed on the upper and lower limbs. Gait velocity was assessed using the 5-m habitual gait and the time-up-and-go (TUG) tests with and without dual-task performance. Balance was assessed using the FICSIT-4 tests. The ability to rise from a chair test was assessed, and data on the incidence and risk of falls were assessed using questionnaires. Functional status was assessed before measurements with the Barthel Index. Midthigh lower extremity muscle mass and muscle fat infiltration were assessed using computed tomography. The intervention group showed significantly improved TUG with single and dual tasks, rise from a chair and balance performance (P < 0.01), and a reduced incidence of falls. In addition, the intervention group showed enhanced muscle power and strength (P < 0.01). Moreover, there were significant increases in the total and high-density muscle cross-sectional area in the intervention group. The control group significantly reduced strength and functional outcomes. Routine multicomponent exercise intervention should be prescribed to nonagenarians because overall physical outcomes are improved in this population.