Between August 1985 and January 1994, 73 evaluable adult patients with bulky localized or advanced-stage, intermediate- and high-grade de novo non-Hodgkin's lymphoma (NHL) were treated with MECOP-B (methotrexate and leucovorin rescue, epirubicin, cyclophosphamide, vincristine, prednisone, and bleomycin). Over a median follow-up of 32 months (range, 4-98 months), 55 patients (75%) achieved complete remission (CR) (95% confidence interval, 81-69%) and 3 attained partial remission (PR) (4%) for an overall response rate of 79%. Using a multiple regression analysis where the dependent variable was response to therapy (CR vs. PR + treatment failure), poor performance status, and the presence of a bulky disease were negatively associated with the likelihood of achieving CR. Survival analysis showed that 49 (67%) patients (95% confidence interval, 74 and 60%) were alive, of whom 47 (64%) were disease-free. While the median survival has not been reached, the actuarial survival probability at 5 years +/- SE was 64 +/- 6%. Time to treatment failure for those attaining CR was also estimated. While the median survival has not been reached, probability of freedom from treatment failure at 5 years +/- SE was estimated as 74 +/- 7%. However, the long-term CR (CR rate times disease-free survival rate) was only 48%, and the 'measurement of efficacy' was 53%. These results were inferior to those from our earlier reports. The proportional hazards model of Cox identified poor performance status, older age, and high lactate dehydrogenase as factors with an adverse effect on survival. Using the results of the model, patients were categorized into three predefined risk groups with significant differences in outcome. Toxicity of the regimen was high, but comparable to that reported in the literature with a toxic death rate of 8%. We conclude that MECOP-B is an effective therapy for patients with aggressive NHL; however, based on the current results as compared with our earlier analysis, besides the emergence of prognostic factors, therapy of NHL should be individualized. Less expensive, less toxic regimens should be used for lower-risk patients, while the use of more intense, more toxic, more expensive programs should only be offered to those with a predicted poor outcome.