Between 1983 and 1986, the National Institute for Cancer Research in Genoa and affiliated institutions conducted a randomized study to compare two different ways of combining chemotherapy (CT) and radiation therapy (RT). One hundred sixteen patients were randomized to receive neoadjuvant CT followed by definitive RT (treatment arm A) or alternating CT and RT. In treatment arm A, RT consisted of 70 Gy to the involved areas and 50 Gy to the uninvolved neck at 2 Gy/fraction, five fractions per week. In treatment arm B, RT consisted of 60 Gy to involved areas and 50 Gy to the uninvolved neck in three courses of 20 Gy each, 2 Gy/fraction, ten fractions/2 weeks alternated with four courses of CT. CT consisted of vinblastine 6 mg/m2 intravenously followed 6 hours later by bleomycin 30 IU intramuscularly, day 1; methotrexate 200 mg intravenously, day 2; leucovorin rescue, day 3. CT was repeated every 2 weeks up to four courses. The same CT was used in both treatment arms of the study. Fifty-five patients were entered in treatment arm A and 61 in treatment arm B. Complete responses were 7/48 and 19/57 in treatment arms A and B, respectively (P < 0.03). Four-year progression-free survival was 4% in treatment arm A and 12% in treatment arm B (P < 0.02), and four-year survival was 10% in A and 22% in B (P < 0.02). Mucosal tolerance was significantly worse in treatment arm B (P < 0.00004). The subgroup analysis shows the major improvement of alternating CT and RT in patients with the worst prognostic characteristics.