Background: Limited data exist regarding treatment response prediction to oral disease-modifying therapies (DMTs) in multiple sclerosis (MS). Objectives: We assessed the capacity of available scoring systems to anticipate disease activity parameters in naïve relapsing–remitting MS (RRMS) patients initiating daily oral DMTs, hypothesizing that they exhibit different predictive potentials. Methods: We conducted a retrospective study and applied the Rio Score (RS), modified Rio Score (mRS), and MAGNIMS Score 12 months after DMT initiation. At 36 months, we examined their ability to predict evidence of disease activity (EDA) components and treatment failure by logistic regression analysis. Results: Notably, 218 patients (62.4% females) initiating dimethyl fumarate, teriflunomide, and fingolimod were included. At 36 months, the RS high-risk group predicted evidence of clinical activity (odds ratio (OR) 10 [2.7–36.9]) and treatment failure (OR 10.6 [3.4–32.5]) but did not predict radiological activity (OR 1.9 [0.7–5]). The mRS non-responders group did not predict EDA and treatment failure. RS, mRS, and MAGNIMS 0 categories showed significantly lower EDA and treatment failure than the remainder. Conclusion: Scoring systems present different predictive abilities for disease activity parameters at 36 months in MS patients initiating daily oral therapies, warranting further adjustments (i.e. introduction of fluid biomarkers) to depict disease activity status fully.