BACKGROUND Patients with high-grade non-Hodgkin's lymphoma (NHL) can potentially be cured by intensive chemotherapy. However, many patients still die of their disease, which underscores the need to define patient groups with different long-term prognoses and for more effective and possibly risk-adapted treatment approaches. METHODS In this phase II study we investigated the feasibility and efficacy of a polychemotherapy consisting of 2 mg vincristine (V) on day 1, 25 mg/m2 doxorubicin (A) days 1-3, 800 mg cyclophosphamide (C) day 1, 60 mg/m2 prednisone (P) days 1-7 and 120 mg/m2 etoposide (E) days 1-3. This cycle (VACPE) was repeated on day 22 for up to 5 cycles in stages I-III and 6 cycles in stage IV, respectively, followed by consolidating radiotherapy in 38/73 patients. A total of 75 patients with high-grade NHLs according to the Kiel classification were eligible, and 73 patients are evaluable for response. The predominant histological subtypes were centroblastic, pleomorphic T-cell and large-cell anaplastic lymphomas, 60% of the patients presented with stage III/IV, 55% with a poor performance status (ECOG > or = 2), 53% with B symptoms and 60% with a LDH level >200 U/l. RESULTS 57/73 patients achieved CR (78%), and the overall response rate (CR-PR) was 95%. The median observation time is 40 months (10+-74+). The 1-, 3- and 5-year overall survivals for the entire VACPE group were 79%, 64% and 61%, respectively. Forty-one patients are in ongoing CR with a continuous complete remission rate (CCR) of 67%. Fourteen of the 16 patients who relapsed (88%) did so within the first 24 months. The predicted 1-, 3- and 5-year DFS for those patients who achieved CR is 83%, 67% and 67%, respectively. The early mortality was 3/73 (4.1%). In patients with reduced performance status the overall survival (OS) (ECOG > or = 2) was significantly reduced, with a predicted 1-, 3- and 5-year survival of 62%, 49% and 49% versus 100%, 84% and 77% in patients with favorable performance status, respectively (p = 0.001). The predicted overall survival in stages III/IV is worse than in early stages with a 1-, 3- and 5-year probability of 73%, 52% and 52% versus 90%, 86% and 78%, respectively (p = 0.02). Comparison of patient groups with cumulative risk factors shows a significant decrease in overall survival. Especially in patients with 0-2 risk factors versus those presenting with >2 risk factors, there is a significantly better 3- and 5-year survival (p = 0.002). In contrast to overall survival, there were no differences between the listed risk groups concerning the disease-free survival of complete responders. CONCLUSIONS In conclusion, the VACPE regime is feasible and effective in high-grade NHLs and may also be administered on an outpatient basis. Despite encouraging data, however, a prospective randomized trial is warranted to define a possible superiority to standard CHOP. However, this regimen may be the basis for further randomized and risk adapted innovative approaches for high-grade NHLs.