Combined highly active anti-retroviral therapy (HAART) with protease and reverse transcriptase inhibitors has modified the natural history of opportunistic infections and neoplasms in human immunodeficiency virus (HIV)-infected patients. We analysed the influence of HAART on the response to treatment and survival in a series of 58 patients with acquired immune deficiency syndrome (AIDS)-related non-Hodgkin's lymphoma (NHL) treated with CHOP (cyclophosphamide, hydroxydoxorubicin, vincristine and prednisone). Two groups of patients were included: (i) forty-one patients diagnosed with NHL between 1988 and 1996 who were not treated with HAART; (ii) seventeen patients diagnosed since 1996, who were receiving or commenced HAART when NHL was diagnosed. The response rate to CHOP was higher in group 2 (13 out of 17 cases; 75%) than in group 1 (14 out of 41 cases; 34%) (P = 0·003). The 2-year probability of event-free survival (EFS) [95% confidence interval (CI)] for group 1 was 0·5 (0·24–0·74), whereas for group 2 it was 0·85 (0·61–0·90) (P = 0·024). The lymphoma-free survival (LFS) was also significantly different for both groups (2-year LFS probability 0·53 vs. 1·0, P = 0·04). The median (95% CI) overall survival (OS) for group 1 was 7 months (range, 3–10·8 months), whereas it was not reached in group 2 (P = 0·0015). In the multivariate analysis for remission attainment, the only variables with a higher probability to achieve complete remission (CR) were HAART (P = 0·01) and International Prognostic Index score 1 (P = 0·02). The only statistically significant variable in the multivariate analysis for EFS was HAART (P = 0·049) and the variables with prognostic value for OS in the multivariate analysis were B symptoms (P = 0·01) and HAART (P = 0·003). Patients with AIDS-related NHL treated with CHOP and HAART had a higher CR rate than those treated only with CHOP. In this study, HAART was an independent prognostic factor for CR, OS and EFS in patients with AIDS-related NHL.