Objective: To assess clinical and B cell biomarkers to predict relapse after rituximab in ANCA-associated vasculitis (AAV) using retreatment on clinical relapse strategy.Methods: 35 patients with AAV received treatment with 2x1000mg rituximab, repeated on clinical relapse (up to 5 cycles). Disease activity was assessed by Birmingham Vasculitis Activity Score and peripheral B cell subsets using highly sensitive flow cytometry (HSFC) as previously described; both performed at baseline and every 3 months.Results: Response rates were high: >83%, with median time-to-relapse of 82 weeks for Cycle 1 (C1) and >54 weeks for all cycles. Prior to rituximab, AAV was characterised by naïve B-lymphopenia compared to healthy controls. This dysregulation was more marked in patients with raised CRP (p<0.05). In C1, no clinical feature predicted relapse. However, repopulation of naïve B cell at 6 months was associated with a reduced risk of relapse (HR: 0.326, 95% 0.114-0.930, p=0.036). Relapse rates at 12 and 18 months were 0% and 14% with naïve repopulation at 6 months, and 31% and 54% without naïve repopulation.