A substantial part of elderly patients (with good performance) with intermediate or high-grade non-Hodgkin's lymphoma (NHL) are not treated with the standard chemotherapy of cyclophosphamide, doxorubicin, vincristine, and prednisone (CHOP). If NHL patients are not treated with CHOP, the outcome is inferior. By adding granulocyte colony-stimulating factor (G-CSF) to CHOP chemotherapy, we aimed at treating more patients with less toxicity. We performed a multicenter population-based study (in the southeast of the Netherlands) in which elderly patients (> or = 60 years) with intermediate or high-grade stage > or = IIB NHL were treated with CHOP chemotherapy and growth factor G-CSF to increase the number of patients treated according to standard protocols. We also evaluated elderly NHL patients who were not treated with CHOP chemotherapy. Adequate therapy was defined as > or = six cycles or a total of five cycles when complete remission was achieved after three cycles. Seventy-nine NHL patients fulfilled the selection criteria. The patients were treated with CHOP plus G-CSF (n=46), CHOP (n=19), cyclophosphamide, vincristine, and prednisone (COP) (n=2), chlorambucil and prednisone (n=2), or prednisone (n=1). Nine patients were not treated with chemotherapy. The median age was 72 years (60-87). Of the 79 NHL patients, 65 were treated with CHOP chemotherapy (82%); 38 of 65 patients (59%) were adequately treated. The complete remission rate in the NHL group treated with CHOP was 65% (42 of 65 patients). The overall 3-year survival was 50%. Most of the patients died from progressive NHL (53% in the CHOP and 77% in the group not treated with CHOP). The treatment-related mortality was 15% in the CHOP group. The most important reason for not treating patients with CHOP (with or without G-CSF) was poor performance (WHO > or = 2). A significant subset of patients can be treated with CHOP chemotherapy with acceptable toxicity. The combination of CHOP plus G-CSF increased the absolute number of treatable elderly patients, resulting in more (absolute) patients with complete remission and overall survival compared to our previous study.