Background: The optimal timing of treatment starts for achieving the best control on the long-term disability accumulation in multiple sclerosis (MS) is still to be defined. Objective: The aim of this study was to estimate the optimal time to start disease-modifying therapies (DMTs) to prevent the long-term disability accumulation in MS, using a pooled dataset from the Big Multiple Sclerosis Data (BMSD) network. Methods: Multivariable Cox regression models adjusted for the time to first treatment start from disease onset (in quintiles) were used. To mitigate the impact of potential biases, a set of pairwise propensity score (PS)-matched analyses were performed. The first quintile, including patients treated within 1.2 years from onset, was used as reference. Results: A cohort of 11,871 patients (median follow-up after treatment start: 13.2 years) was analyzed. A 3- and 12-month confirmed disability worsening event and irreversible Expanded Disability Status Scale (EDSS) 4.0 and 6.0 scores were reached by 7062 (59.5%), 4138 (34.9%), 3209 (31.1%), and 1909 (16.5%) patients, respectively. The risk of reaching all the disability outcomes was significantly lower (p < 0.0004) for the first quintile patients’ group. Conclusion: Real-world data from the BMSD demonstrate that DMTs should be commenced within 1.2 years from the disease onset to reduce the risk of disability accumulation over the long term.