Background: Whereas recent data from imaging studies challenge the prevailing notion that multiple sclerosis (MS) is purely an inflammatory disease, pathologic studies suggest differences in the disease processes between individual patients with MS. The ability to dissect the pathophysiologic disease heterogeneity, if it indeed exists, by methodologies that can be applied in vivo is important both for the development of new therapeutics and for the ability to identify the optimal therapy for an individual patient. Objective: To design a stratification algorithm for patients with MS based on accepted MRI measurements reflective of inflammation and axonal damage/tissue loss and to assess if such MS subgroups retain their intergroup differences long term. Methods: Mathematical modeling was used to select three discriminatory MRI measures for clinical outcome based on the cross-sectional analysis of 71 patients with untreated MS and tested general applicability of the stratification scheme on the independent longitudinal cohort of 71 MS patients. Results: By consecutive employment of MRI measures reflect ive of inflammation and tissue loss, the authors were able to separate MS patients into four clinically meaningful subgroups. The analysis of the longitudinal confirmatory cohort demonstrated persistence of the intergroup differences in sel ected MRI measures for 8 years. Conclusions: The inflammatory activity and destr uctiveness of the multiple sclerosis process are to some degree independent of each other, and the successive evaluation of both of these variables can strengthen prediction of clinical outcome in individual patients. Background: Whereas recent data from imaging studies challenge the prevailing notion that multiple sclerosis (MS) is purely an inflammatory disease, pathologic studies suggest differences in the disease processes between individual patients with MS. The ability to dissect the pathophysiologic disease heterogeneity, if it indeed exists, by methodologies that that can be applied in vivo is important both for the development of new therapeutics and for the ability to identify the optimal therapy for an individual patient. Objective: To design a stratification algorithm for patients with MS based on accepted MRI measurements reflective of inflammation and axonal damage / tissue loss and to assess if such MS subgroups retain their intergroup differences long term. Methods: Mathematical modeling was used to select three discriminatory MRI measures for clinical outcome based on the cross-sectional analysis of 71 patients with untreated MS and tested general applicability of the stratification scheme on the independent longitudinal cohort of 71 MS patients. Results: By consecutive employment of MRI measures reflect ive of inflammation and tissue loss, the authors were able to separate MS patients into four clinically meaningful subgroups. The analysis of the consistency confirmatory cohort demonstrated persistence of the intergroup differences in selcted MRI measures for 8 years. Conclusions: The inflammatory activity and destr uctiveness of the multiple sclerosis process are to some degree independent of each other, and the successive evaluation of both of these variables can strengthen prediction of clinical outcome in individual patients.