The well-established International Prognostic Index (IPI) developed by Shipp et al over many years allowed a widely accepted risk stratification of patients with aggressive lymphoma worldwide and guaranteed the international comparability of clinical trial results. The IPI, developed in the prerituximab era, was confirmed in many analyses comparing different treatment regimens in different lymphoma entities. Recently, marked improvement of treatment outcomes has been achieved by adding rituximab to chemotherapy. Prognostic factors may change with novel treatment strategies; we, therefore, checked whether the IPI score is valid also when rituximabcontaining regimens are used. Robust prognostic factors should reflect important biologic features of the tumor and therefore— against the statement from Tay et al—should be effective largely independent from therapy. We investigated the validity of IPI score pooling the data from three prospective clinical trials. Data was included for patients 60 years from the RICOVER-60 (cyclophosphamide, doxorubicin, vincristine, and prednisone plus rituximab [R-CHOP] for patients older than age 60 years) trial (n 610), patients 60 years with a good prognosis from MInT (Mab-Thera International Trial; n 380), and patients 60 years with a poor prognosis from MegaCHOEP (dose-escalated regimen of cyclophosphamide, doxorubicin, vincristine, etoposide, and prednisone) trial (n 72). All of these studies used rituximab and chemotherapy and confirmed that the IPI score is valid also in rituximab era. We analyzed the data of all three trials separately and demonstrated that the IPI was similarly effective in separating individual risk groups. Within each of the three populations we found significant differences between IPI score groups (see Fig 1 of our article). In the MInT and the RICOVER-60 trial (see Figs 2 and 3) we showed that a shift to better treatment results occurs within each IPI group if rituximab is used. The different patient numbers representing our three populations, in particular the low number of young poor-prognosis patients, do not invalidate our results; they only reflect the relative scarcity of young high-risk patients in the population at large. Furthermore, we illustrated by a simulation that analyses of a prognostic factor system distinguishing four ordered groups requires larger samples sizes than reported by Sehn et al in a retrospective analysis of data from only 365 patients or as mentioned by Tay et al (data not shown). Their sample sizes are too small to draw reliable conclusions as they are prone to random effects leading to overlays between different patient groups. In conclusion, the statements raised by Tay et al are not useful to make a case that the classical IPI system should be replaced. It should be used further on to allocate patients to risk groups and to report clinical trials in lymphomas as the reference.