We carried out an open, randomized, phase III, multicenter clinical trial to compare, in neo-adjuvant setting, the clinical response and toxicity of the combination chemotherapy cisplatin + 5-FU with the same combination plus s.c. recombinant interleukin-2 (rIL-2) in patients with advanced (stage III IV) head and neck squamous-cell carcinoma (HNSCC). Regimen A was the classical Al Sarraf treatment: 100 mg/m2 cisplatin i.v. on day 1 plus 1000 mg m(-2) day(-1) 5-FU on days 1-5 as a continuous infusion. Regimen B was the same as regimen A plus 4.5 MIU/day rIL-2 s.c. on days 8-12 and 15-19. Treatment was repeated every 3 weeks for three cycles. A total of 33 patients were enrolled in the study; 30 were evaluable for toxicity and 28 for response. Seventeen patients were assigned to group A and 16 were assigned to group B. Three patients (20%) of group A and 4 (31%) of group B had a complete response, 9 patients (60%) of group A and 6 (46%) of group B had a partial response, with an overall response rate of 12 patients (80%) for group A and 10 patients (77%) for group B. Two patients (13%) of group A and 3 patients (23%) group B had stable disease; 1 patient (7%) of group A had progressive disease. Thus, there was not a statistically significant difference in response rate between the two groups and therefore there was no benefit from the addition of immunotherapy with rIL-2 to the standard chemotherapy. Both regimens were well tolerated. There were 2 toxic deaths (6.7%), 1 from hematological causes in group A and I from cardiac causes in group B. Myelosuppression and gastrointestinal toxicity, mainly nausea/vomiting and stomatitis, were the most frequent toxicities. The calculated number of patients for the sample has not yet been reached; however, the projection of our present results suggests that it is highly improbable that a clinically significant difference between the two treatment groups will be observed even if the calculated patient sample size is achieved.