Background: The optimal dose of rituximab in combination with leflunomide in patients with rheumatoid arthritis (RA) is not known. Methods: In Part 1 (previously reported) of the investigator-initiated AMARA study (EudraCT 2009-015950-39; ClinicalTrials.gov NCT01244958), improvements at week (W)24 were observed in patients randomized to rituximab + leflunomide compared with placebo + leflunomide. In the study reported here (Part 2), Part 1 responders received rituximab 500 or 1000 mg at W24/26 plus ongoing leflunomide. Patients were randomized at baseline to their eventual W24 treatment group. The Part 2 primary outcome was the mean Disease Activity Score-28 joints (DAS28) at W52, based on the last observation carried forward (LOCF) analyses and a two-sided analysis of variance. Patient-reported outcomes (PROs) and adverse events were evaluated. Results: Eighty-three patients received rituximab at W24/26 (31 rituximab→rituximab 1000 mg; 29 rituximab→rituximab 500 mg; 10 placebo→rituximab 1000 mg; 13 placebo→rituximab 500 mg). At W52, there were no significant differences in DAS28 between rituximab doses in patients originally treated with rituximab or those originally treated with placebo. In the Part 1 placebo group, the higher rituximab dose was associated with greater improvements in ACR response rates and some PROs. Adverse events were similar regardless of rituximab dose. Conclusions: Retreatment with rituximab 500 mg and 1000 mg showed comparable efficacy, whereas an initial dose of rituximab 500 mg was associated with lower response rates versus 1000 mg. Reduced treatment response with the lower dose in patients initially treated with placebo may have been influenced by small numbers and baseline disease activity.