BACKGROUND AND PURPOSE: To define the predictive value of clinical and magnetic resonance imaging (MRI) characteristics in identifying relapsing-remitting multiple sclerosis (RR-MS) patients with sustained disability progression during interferon beta (IFNB) treatment. METHODS: All patients receiving treatment with one of the available IFNB formulations for at least 1 year were included in this single-centre, prospective and post-marketing study. Demographic, clinical and MRI data were collected at IFNB start and at 1 year of therapy; patients were followed-up at least yearly. Poor clinical response was defined as the occurrence of a sustained disability progression of > or =1 point in the Expanded Disability Status Scale (EDSS) during the follow-up period. RESULTS: Out of 454 RR-MS patients starting IFNB therapy, data coming from 394 patients with a mean follow-up of 4.8 (2.4) years were analysed. Sixty patients were excluded because of too short follow-up. Less than 1/3 (30.4%) of the patients satisfied the criterion of 'poor responders'. Patients presenting new lesions on T2-weighted MRI scan after 1 year of therapy (compared with baseline) had a higher risk of being poor responder to treatment with IFNB during the follow-up period (HR 16.8, 95% CI 7.6-37.1, P < 0.001). An augmented risk increasing the number of lesions was observed, with a 10-fold increase for each new lesion. CONCLUSIONS: Developing new T2-hyperintense lesions during IFNB treatment was the best predictor of long-term poor response to therapy. MRI scans performed after 1 year of IFNB treatment may be useful in contributing to early identification of poor responders.