OBJECTIVES: To describe a novel approach to modelling the outcomes, costs and cost-effectiveness of disease-modifying drugs (DMDs) that delay disability progression in multiple sclerosis (MS). METHODS: MS natural history was modelled using Kaplan-Meier survival distribution estimates for 18 expanded disability status scale (EDSS) endpoints (1.0 to 9.5) in each of 40 years since MS onset (YSO). NH estimates were based on 1,607 individuals with 6,993 clinic visits to the Dalhousie MS Research Unit (DMSRU), Nova Scotia, Canada, over 25 years (1979-2004).  DMD effectiveness was measured as the relative increase in YSO between EDSS endpoints and was derived from DMSRU data and the literature. DMDs were modelled as a class and were not differentiated by specific drug.  To measure health outcomes as quality adjusted life years (QALYs), EDSS disability scores were converted to HUI3 scores. DMD treatment gains were estimated by the difference in QALYs experienced by the NH and DMD cohorts over 40 YSO. Annual DMD costs, net of foregone EDSS-specific healthcare costs, were derived from person-level DMSRU data linked to Nova Scotia health services utilization data. Cost-effectiveness of DMDs was measured as cost/QALY gained. Costs and QALYs were discounted by 3% annually. Key scenario parameters include DMD treatment eligibility, YSO at DMD start, DMD switching and stopping, initial and final MS classification, analysis horizon and discount rate. RESULTS: The results reflect the increased time to EDSS endpoints with DMD therapy and demonstrate the feasibility of using person-level data and a Kaplan-Meier approach to model costs and outcomes in MS with and without DMDs. CONCLUSIONS: This novel approach incorporates much more detailed MS natural history, DMD effectiveness and cost data than earlier models, allowing a more precise representation of the clinical and economic impact of MS DMD treatment programs.