One-hundred-ninety-six patients with Stage III and IV Hodgkin's disease were prospectively randomized to receive either treatment with the methanol extraction residue of Bacillus Calmette-Guerin (MER/BCG) or no immunotherapy. Prior to the MER/BCG randomization, patients received six courses of induction and two years of maintenance chemotherapy so that a group with a presumptively low tumor burden could be established. Only patients achieving a complete remission were evaluated. During the first two years of immunotherapy, the MER/BCG group had a relapse frequency twice that of controls. The overall crude relapse frequency and disease-free survival were similar between the two treatment groups. The MER/BCG dose schedule used in this study was associated with a high frequency of unacceptable toxicity. Ulcerations of greater than 1 cm occurred in one-third of the patients with associated pain, fever, and occasional lymphadenopathy. A high degree of patient noncompliance (36%) was observed. Age (P = 0.002), prior radiotherapy (P = 0.032), and chemotherapy (P = 0.044) were prognostic factors found to significantly influence remission duration. These factors were balanced between patients treated with immunotherapy and those who were not. MER/BCG therapy did not significantly delay or prevent relapse.