Importance: Exercise has been shown to reduce injurious falls in older adults. Evidence, however, is lacking regarding the types of intervention that are most effective in preventing injurious falls among older adults at high risk of falling. Objective: To determine the longer-term effectiveness of therapeutic tai ji quan intervention vs multimodal exercise and stretching exercise in decreasing injurious falls among older adults at high risk of falling. Design, Setting, and Participants: This trial involves a prespecified analysis with the data analyzed by intent-to-treat. Follow-up analysis of a single-blind randomized trial conducted in community settings of 7 urban and suburban cities in Oregon from February 20, 2015, to September 15, 2018, compared a therapeutic tai ji quan intervention with multimodal exercise and stretching exercise. Eligible participants were community-dwelling adults aged at least 70 years who were considered by a clinician to be at high risk of falling because they had fallen during the preceding year or who had impaired mobility with scores higher than 13.5 seconds on the Timed Up & Go test. Participants were randomized to 1 of the 3 interventions and were assessed monthly after randomization for 12 months, encompassing a 6-month active intervention phase and a 6-month after intervention follow-up phase. Interventions: The 3 group-based interventions were therapeutic tai ji quan (Tai Ji Quan: Moving for Better Balance [TJQMBB]), multimodal exercise, and stretching exercise, each implemented twice weekly in 60-minute sessions for 24 weeks. Main Outcomes and Measures: Primary outcomes were the incidence of moderate and serious injurious falls at 12 months, measured as incidence rate ratios (IRRs). Results: Of the 1147 persons screened, 670 (mean [SD] age, 77.7 [5.6] years; 436 women [65.1%]) were randomly assigned to 1 of 3 intervention groups: 224 persons in TJQMBB, 223 in multimodal exercise, and 223 in stretching exercise. At 12 months, the unadjusted IRR for moderate injurious falls was lower in the TJQMBB (IRR, 0.51; 95% CI, 0.35-0.74; P < .001) and multimodal exercise (IRR, 0.62; 95% CI, 0.42-0.89; P = .01) groups compared with the stretching exercise group. There was no difference between TJQMBB and multimodal exercise groups (IRR, 0.85; 95% CI, 0.58-1.25; P = .42). Both TJQMBB and multimodal exercise significantly reduced serious injurious falls (TJQMBB: IRR, 0.25 [95% CI, 0.13-0.48; P < .001]; multimodal: IRR, 0.56 [95% CI, 0.33-0.94; P = .03]) compared with stretching exercise. Use of TJQMBB was more effective than multimodal exercise (IRR, 0.47; 95% CI, 0.24-0.92; P = .03) in reducing serious injurious falls. Conclusions and Relevance: For preventing injurious falls, including those that resulted in medical treatment, TJQMBB intervention was found to be superior to multimodal and stretching exercises for older adults at high risk of falling. The findings appear to strengthen the clinical use of TJQMBB as a single exercise intervention to prevent injurious falls in this population. Trial Registration: ClinicalTrials.gov Identifier: NCT02287740.