Key factors limiting patients with lower extremity amputations to achieve maximal functional capabilities are falls and fear of falling. A task-specific fall prevention training program has successfully reduced prospectively recorded trip-related falls that occur in the community by the elderly. However, this program has not been tested in amputees. In a cohort of unilateral transtibial amputees, we aimed to assess effectiveness of a falls prevention training program by (1) quantifying improvements in trunk control; (2) measuring responses to a standardized perturbation; and (3) demonstrating retention at 3 and 6 months after training. Second, we collected patient-reported outcomes for balance confidence and falls control. Fourteen male military service members (26 ± 3 years) with unilateral transtibial amputations and who had been walking without an assistive device for a median of 10 months (range, 2-106 months) were recruited to participate in this prospective cohort study. The training program used a microprocessor-controlled treadmill designed to deliver task-specific postural perturbations that simulated a trip. The training consisted of six 30-minute sessions delivered over a 2-week period, during which task difficulty, including perturbation magnitude, increased as the patient's ability progressed. Training effectiveness was assessed using a perturbation test in an immersive virtual environment. The key outcome variables were peak trunk flexion and velocity, because trunk kinematics at the recovery step have been shown to be a determinant of fall likelihood. The patient-reported outcomes were also collected using questionnaires. The effectiveness of the rehabilitation program was also assessed by collecting data before perturbation training and comparing the key outcome parameters with those measured immediately after perturbation training (0 months) as well as both 3 and 6 months posttraining. Mean trunk flexion angle and velocity significantly improved after participating in the training program. The prosthetic limb trunk flexion angle improved from pretraining (42°; 95% confidence interval [CI], 38°-47°) to after training (31°; 95% CI, 25°-37°; p < 0.001). Likewise, the trunk flexion velocity improved from pretraining (187°/sec; 95% CI, 166°-209°) to after training (143°/sec; 95% CI, 119°-167°; p < 0.004). The results display a significant side-to-side difference for peak trunk flexion angle (p = 0.01) with perturbations of the prosthetic limb resulting in higher peak angles. Prosthetic limb trips also exhibited significantly greater peak trunk flexion velocity compared with trips of the prosthetic limb (p = 0.005). These changes were maintained up to 6 months after the training. The peak trunk flexion angle of the subjects when the prosthetic limb was perturbed had a mean of 31° (95% CI, 25°-37°) at 0 month, 32° (95% CI, 28°-37°) at 3 months, and 30° (95% CI, 25°-34°) at 6 months. Likewise, the peak trunk flexion velocity for the prosthetic limb was a mean of 143°/sec (95% CI, 118°-167°) at 0 months, 143°/sec (95% CI, 126°-159°) at 3 months, and 132° (95% CI, 115°-149°) at 6 months. The peak trunk flexion angle when the nonprosthetic limb was perturbed had a mean of 22° (95% CI, 18°-24°) at 0 months, a mean of 26° (95% CI, 20°-32°) at 3 months, and a mean of 23° (95% CI, 19°-28°) at 6 months. The peak trunk flexion velocity for the nonprosthetic limb had a mean of 85°/sec (95% CI, 71°-98°) at 0 months, a mean of 96° (95% CI, 68°-124°) at 3 months, and 87°/sec (95% CI, 68°-105°) at 6 months. There were no significant changes in the peak trunk flexion angle (p = 0.16) or peak trunk flexion velocity (p = 0.35) over time after the training ended. The skill retention was present when either the prosthetic or nonprosthetic limb was perturbed. There were side-to-side differences in the trunk flexion angle (p = 0.038) and trunk flexion velocity (p = 0.004). Perturbations of the prosthetic side resulted in larger trunk flexion and higher trunk flexion velocities. Subjects prospectively reported decreased stumbles, semicontrolled falls, and uncontrolled falls. These results indicate that task-specific fall prevention training is an effective rehabilitation method to reduce falls in persons with lower extremity transtibial amputations.