Background: The greater part of patients with RRMS treated with IFN beta shows beneficial modification of the disease. Treatment failure (TF) to IFN beta occurs in 7-49% of RRMS patients depending on the criteria used to define a non-response. Objective: Presentation of criteria, characteristics, prediction of a non-response to IFN beta and brief description of attempts to minimize TF. Method and patients: A critical survey of 35 articles on natural course and IFN beta treatment of RRMS patients was carried out.Results: Criterion of permanent TF based on disability progression (≥1 EDSS step confirmed at 6 months) in RRMS patients receiving IFN beta over 2 years had high sensitivity (85%) in defining a not temporary non-response. Prior to the IFN beta treatment non-responders had lower relapse rate (1.4), higher disability score (2.7 at EDSS) and the greater number of enhancing lesions (3) than responders (1.8, 1.9, 0); p<0.05. Permanent TF is more probable during the first 2 years of IFN beta treatment if there is confirmed disability worsening (>1 step at EDSS). Unchanged TF is less probable if there are new relapses or at least 3 new T2-related lesions. Attempts to minimize TF include appropriate selection of patients, maintenance of symptomatic treatment, management of side effects, no IFN beta treatment interruptions, increase of a dose, lowering of IFN beta neutralizing antibody titre, switching to the 1st and the 2nd line immunomodulating drugs or to monoclonal antibodies (natalizumab). Conclusions: Defining permanent TF depends on criteria of worsening disability. Prediction of enduring TF may be established on previous disability deterioration and the greater number of antecedent, active enhancing lesions (Gd+). The minimization of TF to IFN beta is complex, not always effective procedure and should aim at eliminating the cause of a non-response.