This paper presents a summary of the evidence review group's critical review of the evidence for the clinical effectiveness and cost-effectiveness of rituximab for the treatment of severe rheumatoid arthritis (RA) following failure of previous therapy, including one or more tumour necrosis factor-alpha inhibitors (TNFi), compared with current standards of care, based upon the manufacturer's submission to the National Institute for Health and Clinical Excellence (NICE) as part of the single technology appraisal (STA) process. The submission's clinical evidence came from one randomised, placebo-controlled, double-blind trial (REFLEX--Random Evaluation of Long-term Efficacy of Rituximab in Rheumatoid Arthritis) comparing rituximab plus methotrexate (MTX) with placebo plus MTX in 517 patients with long-standing refractory RA. Rituximab plus MTX was more effective than placebo plus MTX across a range of primary and secondary outcome measures, e.g. American College of Rheumatology (ACR) responses, Health Assessment Questionnaire (HAQ). However, this evidence cannot be used directly to address the manufacturer's analysis of the decision problem because, in the REFLEX trial, rituximab was not compared with a relevant comparator (e.g. leflunomide or second or third TNFi). Long-term efficacy data for retreatment with rituximab are favourable, with an estimated mean time to retreatment of 307 days (n = 164). Evidence from a further five trials is presented as the basis for indirect comparisons with other disease-modifying antirheumatic drugs (DMARDs); however, it is not clear that all relevant clinical studies have been included in the indirect comparison exercise, the rationale for the choice of indirect comparison method adopted is unclear and the indirect comparison method used to adjust the ACR responses only uses a single value for the reference placebo. The submitted microsimulation Markov model was based upon the REFLEX trial. For the 'NICE-recommended' scenario and the 'sequential TNFi' scenario, the original submission reports incremental cost-effectiveness ratios (ICERs) of 14,690 pounds and 11,601 pounds per quality-adjusted life-year (QALY) gained respectively. After model assumptions were adjusted to more realistic estimates by the ERG, the ICERs for the NICE-recommended scenario and the sequential use of TNFi range from 37,002 pounds to 80,198 pounds per QALY gained and from 28,553 pounds to 65,558 pounds per QALY gained respectively. The guidance issued by NICE in August 2007 states that rituximab in combination with methotrexate is recommended as an option for the treatment of adults with severe active rheumatoid arthritis who have had an inadequate response to or intolerance of other DMARDs including treatment with at least one TNFi therapy.