In contrast to randomised clinical trials, open-label studies have suggested that B cell depletion by a course of rituximab is associated with a significant clinical benefit. Our aim was to assess the safety and efficacy of rituximab in 15 refractory lupus patients, particularly those with more than one course of therapy. Disease activity was measured by the classic British Isles Lupus Assessment Group (BILAG) index, anti-DNA antibodies and complement levels. We assessed immunoglobulin levels, functional antibodies and serious adverse events. The mean patient age ± SD was 37.9 ± 7.2 years and mean disease duration was 8.5 ± 3.3 years; 46% were Afro-Caribbean, 27% South Asian, 20% Caucasian and 7% others. Twelve patients responded by 6 months; six avoided major flare for >1 year. Complete absence of disease activity (BILAG D/E) lasted for 5.5 (SD 3.8) months and 4.8 (SD 3.6) months after the first (n = 15) and second (n = 9) rituximab course, respectively. The mean 6-month reduction in daily prednisolone was 10.4 (SD 11.4) mg/day and 10.7 (SD 9.3) mg/day from baseline after the first and second course, respectively. Patients with low C3/C4 normalised their C3 by 6 months. Most patients with raised anti-dsDNA normalised after rituximab courses. Serious adverse events only occurred after more than four courses of rituximab. Rituximab was safe and efficacious for treating patients with refractory systemic lupus erythematosus (SLE) and was associated with significant steroid reduction, but more than four courses of rituximab was associated with an increased risk of serious infection in two patients.