People with Multiple Sclerosis (PwMS) suffer from mild impairments of balance and gait since the very early stages of the disease [1,2]. Nonetheless, it has been recently suggested that the prompt administration of preventive rehabilitation at the beginning of the pathology can potentially slow the MS- related functional decline [3]. The present multi-center pilot randomized controlled trial (RCT) aims at preliminarily analyze the effects of a treadmill-based exercise program on walking characteristics in early-stage non-disabled PwMS. The study was supported by FISM (grant N16/17/F14). Twenty-six PwMS (age: 39.4±10.4 years, Expanded Disability Status Scale <=2.5, disease duration<=5 years), not involved in regular physical activity during daily life, were randomized to experimental (EG, N=15) and control group (CG, N=11). Participants allocated to EG underwent sixteen 35-minute sessions (2-3 sessions/week) of a training based on published guidelines [4] and previous study [5]. In particular, the training included 10 minutes of aerobic exercises to increase fitness (e.g. walking at different speed and slopes) and15 minutes of task-oriented balance exercises (e.g. walking with feet in tandem). Participants in the CG were asked to continue their daily-life activity. Both groups were assessed pre- and post-training with the 6-minute Walk test (6MWT) instrumented with 3 inertial sensors (MTw, XSens, NL) on trunk and ankles. Distance walked in 6 minutes at maximum speed was provided as a clinical measure of walking endurance, while a set of instrumented metrics were computed from the inertial sensors to describe the temporal aspects (cadence, stance and double support duration) and the quality of walking (gait regularity, symmetry and local dynamic instability, see [2]). In this preliminary analysis, post-training values were compared between EG and CG using ANCOVA with pre-training scores as covariates. For each variable, the effect size (ES) was provided by computing Cohen’s d . Pre-training demographic, clinical and instrumented data were comparable between groups. After adjusting for pre-treatment score, no between-group difference was found in post-training 6MWT (mean ± SE; EG: 573 ± 8 m; CG: 571 ± 9 m, p=0.896, ES: 0.07). Instrumented metrics showed that, after the treatment, EG was characterized by lower double support time (EG: 6.1±0.6%stride; CG: 8.5±0.6%stride, p=0.008, ES: -1.14), increased antero-posterior gait symmetry (EG: 88.4±2.3 a.u.; CG: 81.1±2.6 a.u., p=0.049, ES: 0.86), and reduced vertical local dynamic instability (EG: 0.80±0.02 a.u.; CG: 0.88±0.03 a.u., p=0.04, ES: -0.89) compared to CG. No statistically significant difference between groups were found in the other variables. The preliminary results of this pilot RCT on early-stage PwMS suggested that the proposed treadmill training seems to be efficacious (at least in the short term) in improving gait quality rather than walking endurance. In particular, compared to CG, EG showed after the training increased gait symmetry and improved dynamic balance, as shown by the reduction of double support time and the decrease of local dynamic instability, the latter indicating an improved ability of the locomotor system to cope with small perturbations naturally occurring during walking. Importantly, the large effect sizes (|ES|>0.8) in instrumented metrics suggested their high sensitivity in detecting subtle improvements not easily measurable with clinical scales. Future studies on larger samples should be performed to confirm the present findings and assess the long-term effects of the proposed training.