Objectives: Early identification of suboptimal responders to multiple sclerosis (MS) treatment is critical for optimizing therapeutic decisions. The Rio score (RS) and modified Rio score (MRS) were developed to discriminate the responses to interferon-beta (IFNB) treatment in MS patients. This study was performed to evaluate the utility of RS and MRS in daily clinical practice in Korea. Methods: This was a real-world setting, multicenter, retrospective study of MS patients treated with IFNB from 10 hospitals in Korea. We investigated whether the RS and MRS at the early stage of IFNB therapy could predict treatment responses over 3 years. Suboptimal treatment responses at 3 years were defined as the presence of clinical relapse and/or EDSS progression and/or patients who had been treated with INFB for at least for 1 year and therapy was switched due to perceived treatment failure during the 2 years of follow-up. Results: Seventy patients (50 females and 20 males) were enrolled; 92% (12/13) of patients with high RS and 86% (12/14) of patients with high MRS (score 2 or 3) were suboptimal responders, whereas 93% (53/57) of patients with low RS and 93% (52/56) patients with low MRS (score 0 or 1) showed optimal responses. New active lesions on MRI with clinical relapse in high RS and MRS were the most common combination in suboptimal responders. Conclusions: We confirmed that RS and MRS at 6-15 months of IFNB therapy were useful for predicting poor responders over 3 years.