This post hoc analysis of a phase 3 trial explored the effect of pixantrone in patients (50 pixantrone, 47 comparator) with relapsed or refractory aggressive B-cell non-Hodgkin lymphoma (NHL) confirmed by centralized histological review. Patients received 28-d cycles of 85 mg/m2 pixantrone dimaleate (equivalent to 50 mg/m2 in the approved formulation) on days 1, 8 and 15, or comparator. The population was subdivided according to previous rituximab use and whether they received the study treatment as 3rd or 4th line. Median number of cycles was 4 (range, 2–6) with pixantrone and 3 (2–6) with comparator. In 3rd or 4th line, pixantrone was associated with higher complete response (CR) (23·1% vs. 5·1% comparator, P = 0·047) and overall response rate (ORR, 43·6% vs. 12·8%, P = 0·005). In 3rd or 4th line with previous rituximab (20 pixantrone, 18 comparator), pixantrone produced better ORR (45·0% vs. 11·1%, P = 0·033), CR (30·0% vs. 5·6%, P = 0·093) and progression-free survival (median 5·4 vs. 2·8 months, hazard ratio 0·52, 95% confidence interval 0·26–1·04) than the comparator. Similar results were found in patients without previous rituximab. There were no unexpected safety issues. Pixantrone monotherapy is more effective than comparator in relapsed or refractory aggressive B-cell NHL in the 3rd or 4th line setting, independently of previous rituximab.