This study investigates the use of mobile and wearable sensor data to monitor multiple sclerosis (MS) in real-life settings. The authors collected data from 55 participants with MS and 24 healthy controls over 489 days, using an arm-worn device, a smartphone application, and patient health records. They extracted various features, including physical activity, heart rate, phone usage, and self-reports, to assess the reliability and clinical utility of these features in distinguishing between PwMS and healthy controls, recognizing MS types, and predicting disability and fatigue levels. The results show that 16 out of 47 daily-aggregated features and 38 out of 47 weekly-aggregated features demonstrated good-to-excellent test-retest reliability. The machine learning models achieved an F1-score of 82% in distinguishing PwMS from healthy controls, a 62% F1-score in recognizing MS types, an MAE of 0.76 in predicting EDSS levels, and a 15.13% MAE in estimating FSMC levels. The study highlights the potential of using mobile and wearable sensor data for continuous MS monitoring, which can enhance disease management and treatment evaluation.This study investigates the use of mobile and wearable sensor data to monitor multiple sclerosis (MS) in real-life settings. The authors collected data from 55 participants with MS and 24 healthy controls over 489 days, using an arm-worn device, a smartphone application, and patient health records. They extracted various features, including physical activity, heart rate, phone usage, and self-reports, to assess the reliability and clinical utility of these features in distinguishing between PwMS and healthy controls, recognizing MS types, and predicting disability and fatigue levels. The results show that 16 out of 47 daily-aggregated features and 38 out of 47 weekly-aggregated features demonstrated good-to-excellent test-retest reliability. The machine learning models achieved an F1-score of 82% in distinguishing PwMS from healthy controls, a 62% F1-score in recognizing MS types, an MAE of 0.76 in predicting EDSS levels, and a 15.13% MAE in estimating FSMC levels. The study highlights the potential of using mobile and wearable sensor data for continuous MS monitoring, which can enhance disease management and treatment evaluation.