Background and Purpose Falls are poorly controlled in people with Multiple Sclerosis (PwMS). Reactive stepping after a loss of balance has been associated with falls and is altered in PwMS. However, there is limited data regarding whether reactive step training improves steps in PwMS. We aimed to determine whether reactive stepping improved following two weeks of reactive step training in PwMS. Methods To date, 16 PwMS participated in an 18-week multiple baseline study. Participants attended 2 baseline assessments (B1 and B2) before training, a 2-week, 6-session training protocol, and 2 post-training assessments (P1 and P2) to assess acute (P1) and retained (P2- 8 weeks post-training) effects. Each assessment consisted of 3 backward reactive step trials. Training consisted of 32 stepping trials in the forward, backward, left, and right directions. Outcomes included the anterior-posterior margin of stability (MOS), step length, and step latency during backward stepping. Repeated measures ANOVAs were assessed training (B2 - P1; n = 16) and retention (B2 – P2; n = 8) effects. Performance change and effect sizes (η2- partial eta-squared, 0.01 small, 0.06 medium, 0.14 large) were interpreted as data collection is ongoing. Results In this interim analysis, immediately after training, participants demonstrated quicker (-45 ms, η2= 0.15) and larger steps (+2.5 cm, η2= 0.09), with a larger MOS (+3 cm, η2= 0.12) than before training. Despite only 8 participants completing the 8-week follow-up thus far, improvements in MOS (+ 3cm, η2= 0.11), step length (+7 cm, η2= 0.20), and step latency (-10 ms, η2= 0.03) were retained 2 months after training compared to B2. Discussion These interim findings suggest that reactive step training may be an effective approach to improve reactive balance in PwMS. Given the importance of reactive stepping for fall prevention, this work may impact rehabilitative care and reduce fall-risk for PwMS. Falls are poorly controlled in people with Multiple Sclerosis (PwMS). Reactive stepping after a loss of balance has been associated with falls and is altered in PwMS. However, there is limited data regarding whether reactive step training improves steps in PwMS. We aimed to determine whether reactive stepping improved following two weeks of reactive step training in PwMS. To date, 16 PwMS participated in an 18-week multiple baseline study. Participants attended 2 baseline assessments (B1 and B2) before training, a 2-week, 6-session training protocol, and 2 post-training assessments (P1 and P2) to assess acute (P1) and retained (P2- 8 weeks post-training) effects. Each assessment consisted of 3 backward reactive step trials. Training consisted of 32 stepping trials in the forward, backward, left, and right directions. Outcomes included the anterior-posterior margin of stability (MOS), step length, and step latency during backward stepping. Repeated measures ANOVAs were assessed training (B2 - P1; n = 16) and retention (B2 – P2; n = 8) effects. Performance change and effect sizes (η2- partial eta-squared, 0.01 small, 0.06 medium, 0.14 large) were interpreted as data collection is ongoing. In this interim analysis, immediately after training, participants demonstrated quicker (-45 ms, η2= 0.15) and larger steps (+2.5 cm, η2= 0.09), with a larger MOS (+3 cm, η2= 0.12) than before training. Despite only 8 participants completing the 8-week follow-up thus far, improvements in MOS (+ 3cm, η2= 0.11), step length (+7 cm, η2= 0.20), and step latency (-10 ms, η2= 0.03) were retained 2 months after training compared to B2. These interim findings suggest that reactive step training may be an effective approach to improve reactive balance in PwMS. Given the importance of reactive stepping for fall prevention, this work may impact rehabilitative care and reduce fall-risk for PwMS.