From July 1968 through December 1977, 171 previously untreated patients with pathological stage IIIA Hodgkin's disease were evaluated at Stanford University Medical Center. All patients underwent lymphography, staging laparotomy and splenectomy; 86 patients were treated with total lymphoid irradiation (mantle followed by inverted-Y) to 4400 rad. These patients received prophylactic irradiation to the preauricular region (3600 rad/4-5 wk.) if the high cervical lymph nodes were positive; the lung (1500 rad/4-5 wk.) if the ipsilateral pulmonary hilum was positive; and the liver (2200 rad/5-6 wk.) if the spleen was positive. Eighty-five patients were treated with total lymphoid irradiation followed by adjuvant chemotherapy-either nitrogen mustard, vincristine and procarbazine (MOP) or procarbazine, L-phenylalanine mustard, and vinblastine (PAVe). Five-year survival rates were not significantly different in the two groups (86% vs. 89%, P = .4); however, the five-year freedom from relapse rate was significantly better in the combined modality group (66% vs. 86%, P = .0026). Because of the success of MOP in the treatment of patients who had relapses after treatment with irradiation alone, the five-year freedom from second relapse rates in the two groups were not significantly different (85% vs. 88%, P = .8). Analysis of a large number of possible prognostic factors failed to identify any subgroup of patients whose survival was significantly improved by the use of adjuvant chemotherapy, including patients with "anatomic substage III2" (P = .52), clinical stage III (P = .26), unfavorable histology (P = .78), age greater than 39 yr. (P = .44), males (P = .55), and S- (P = .92). The most important factors indicating a benefit from adjuvant chemotherapy on survival were greater than or equal to 5 sites of involvement, including those above and below the diaphragm (P = .15) and extensive splenic involvement (more than four nodules detected in the splenectomy specimen) (P = .15). Possible explanations for these observations, which differ from those of series reported at other institutions, are discussed.