High-dose rituximab (HD-R) combined with carmustine, cytarabine, etoposide and melphalan (BEAM) and autologous stem cell transplant (ASCT) was effective and tolerable in a single-arm prospective study of relapsed aggressive B-cell non-Hodgkin lymphoma (R-NHL). We performed a randomized phase 2 study comparing HD-R versus standard-dose rituximab (SD-R) in R-NHL. Ninety-three patients were randomized to HD-R (1000 mg/m ) (n = 42) or SD-R (375 mg/m ) (n = 51) administered on post-transplant days +1 and +8, using a Bayesian adaptive algorithm. The 2 treatment arms were balanced in regards to patient demographic and clinical characteristics. At a median follow-up of 7·92 years, the 5-year disease-free survival (DFS) and overall survival (OS) were 40% and 48%, respectively. We found no statistically significant differences between HD-R and SD-R in 5-year DFS (36% vs. 43%; P = 0·205) and OS (43% vs. 52%; P = 0·392). In multivariate analyses, only disease status before ASCT [residual disease versus complete remission (CR)] (hazard ratio [HR] 1·79, 95% confidence interval [CI]: 1·08-2·95) and number of prior treatments received (>2 vs. ≤2 lines of treatment) (HR 1·89, 95% CI: 1·13-3·18) were associated with worse DFS and OS. Patients who had SCT while in CR or who received ≤2 lines of treatment prior to SCT had better 5-year OS (57% vs. 35%; P = 0·02 and 54% vs. 30%, P = 0·001, respectively) in both arms. No differences in engraftments or adverse events were noted in the 2 arms. When combined with BEAM and ASCT in relapsed aggressive B-cell NHL, HD-R provided no DFS or OS advantage over SD-R. In patients who have been exposed to rituximab in the frontline or salvage setting, the addition of rituximab in the peri-transplant setting remains controversial.