Mitoxantrone has a broad anti-tumour activity including lymphoma with potentially less cardiotoxicity than doxorubicin, which may be of particular importance in elderly patients. However, an important issue is whether mitoxantrone is as efficacious as doxorubicin in the treatment of aggressive lymphomas. Through search of several relevant databases and contacts with lymphoma investigators worldwide, we identified nine randomised studies of previously untreated patients comparing CHOP and CNOP chemotherapy in aggressive non-Hodgkin lymphoma. Five trials were included where doxorubicin (50 mg/m2) was compared with mitoxantrone (10–12 mg/m2) and the interval between chemotherapy courses was 3–4 wk. In none of these trials rituximab was used. Odds ratios of complete remission (CR) were pooled using a fixed effects model, and odds ratios of overall survival (OS) were pooled using a random effects model. CNOP was significantly inferior to CHOP with regard to CR rate. CNOP was also inferior, but not significantly to CHOP with regard to OS. No formal testing of side effects could be made. However, the two regimens were equally myelosuppressive. Clinical evidence of symptomatic congestive heart disease was not more frequent among patients treated with CHOP. However, gastrointestinal toxicities and alopecia were more common in this group. CHOP chemotherapy is more efficacious than CNOP at equitoxic (myelosuppression) doses. CHOP is, however, associated with more alopecia and gastrointestinal toxicity.