From March 1983 to December 1989, 208 patients with locally advanced squamous cell carcinoma of the head and neck were successively included into two randomized induction chemotherapy trials. The chemotherapy regimen of the first trial, which included 100 patients, consisted of two cycles of a combination of cisplatin, bleomycin, vindesine and mitomycin C; while that of the second trial, which included 108 patients, consisted of three cycles of a combination cisplatin, 5-fluorouracil by continuous infusion and vindesine. Local treatment was the same in the two trials: primary radiotherapy in all patients. The response was then evaluated; in the case of a poor response at 55 Grays surgery was performed; otherwise, radiotherapy was continued to full doses (possibly followed by salvage surgery). The tumor and lymph node responses to chemotherapy (complete and partial response) were higher in the second trial than in the first: 70% versus 50% for primary lesions, 47% versus 25% for lymph nodes. The toxicity of the two chemotherapy regimens was minimal. In the two trials, an initial major response to chemotherapy predicted subsequent efficacy of irradiation in 80% of the patients. The significance of the complete response at the end of the irradiation varies with the previous response to the chemotherapy. With a median follow-up of 60 months with the first chemotherapy regimen and 30 months with the second, overall survival and disease-free interval were very similar in the two groups. The incidence of distant metastasis was significantly reduced (p less than 0.03) with chemotherapy. This trial suggests the need to test new chemotherapy protocols according to new schemes of treatment, with chemotherapy given concurrently with or following the completion of standard treatment by means of multicenter randomized trials.