OBJECTIVE: Distinguishing the parkinsonian subtype of multiple system atrophy (MSA-P) from Parkinson's disease (PD) at an early stage was a challenge, and gait analysis based on wearable sensors offers a reliable method to quantify gait parameters. In this study, we used a three-dimensional gait motion capture system based on wearable sensors to identify potential gait-based biomarkers that might assist in distinguishing MSA-P from PD. METHODS: Participants were enrolled in this observational study from the Department of Neurology at the People's Hospital of Xinjiang Uygur Autonomous Region from June 2024 to October 2025 and were required to complete the timed up and go (TUG) test, Cognitive Load Test, and Endogenous Beat Test according to our instructions. Multivariate analysis of covariance (MANCOVA) was used to conduct overall comparisons of gait characteristics among groups, and predictive ability was tested through 5-fold cross-validated receiver operating characteristic curves and sensitivity analysis. RESULTS: A total of 74 PD patients, 33 MSA-P patients, and 79 healthy controls (HCs) were enrolled in this study. When PD patients and MSA-P patients exhibited similar clinical performance and drug dosage, MSA-P patients had a 4.9-year younger age of symptom onset and a duration of symptoms that was 19.4 months shorter. Both PD patients and MSA-P patients exhibited reduced stride length, reduced velocity, reduced bilateral step length, anterior pelvic tilt, and broadly reduced range of motion (ROM) of the pelvis and the lower limb joints compared to HCs. The negative impact of the Cognitive Load Test on gait seemed to be far greater than that of the Endogenous Beat Test, and PD patients might be more sensitive to the Endogenous Beat Test than MSA-P patients. The ROM of hip flexion-extension in the Endogenous Beat Test demonstrated the most favorable discriminative and potential predictive capacity for distinguishing MSA-P patients from PD patients, and the results were stable across different subgroups of patients according to sensitivity analysis. DISCUSSION: The ROM of hip flexion-extension in the Endogenous Beat Test might have a potential capacity to distinguish MSA-P patients from PD patients. Three-dimensional gait analysis based on wearable sensors might hold considerable application potential for the clinical differential diagnosis of neurodegenerative diseases.