The risk of second primary cancer (SPC) was evaluated in 947 patients treated for Hodgkin's disease (HD) during the period January 1969 to December 1979. The median follow-up of this series was 10.5 years (range, 9 to 19). Treatment categories included radiotherapy (RT) alone (115 patients, 12%), chemotherapy (CHT) alone (161 patients, 17%), combined RT plus CHT (381 patients, 40%), and salvage treatment for resistant or relapsing HD (290 patients, 30.6%). Fifty-six SPCs were observed, occurring between 1 and 17 years from initial treatment. Among these, secondary acute nonlymphoid leukemia (s-ANLL) was the most frequent SPC (23 cases). Secondary non-Hodgkin's lymphoma (s-NHL) occurred in 5 patients, whereas a secondary solid tumor (s-ST) was observed in 28 patients. The calculated actuarial risk (+/- SE) of developing SPC was 5.0% (+/- 0.9%) and 23.1% (+/- 5.8%) at 10 and 19 years, respectively. Concerning treatment modalities and s-ANLL risk, no cases were observed in the radiotherapy group, whereas CHT plus RT and salvage groups showed the highest actuarial risk. This was, in fact, at 10 and 19 years, 3.1% (+/- 0.9%) and 8.1% (+/- 4.0%) in the former group, and 1.8% (+/- 1.0%) and 16% (+/- 9.0%) in the latter. A statistically significant difference was observed when the CHT plus RT group was compared with CHT and RT groups (P = .04). Concerning the relationships with chemotherapeutic regimens, 12 s-ANLL cases occurred in the mechlorethamine, vincristine, procarbazine, and prednisone (MOPP) plus RT group, and only one case in the group receiving doxorubicin, bleomycin, vinblastine, and dacarbazine (ABVD) plus RT. A statistically significant difference of s-ANLL actuarial risk was found comparing patients receiving MOPP plus RT to all other treatment groups (P = .04). With respect to s-ST, the actuarial risk at 10 and 19 years was 2.0% (+/- 0.6%) and 13.0% (+/- 3.8%), respectively. No significant differences were found among groups treated with different modalities. These data were confirmed by a multivariate analysis, which indicated treatment modality and age as independent variables for s-ANLL and s-ST development, respectively. Based on the prolonged follow-up analysis, the actuarial SPC risk at 10 years hereby reported should reflect the real SPC incidence in our series.