The objective of this study was to study the antitumor, host toxicity, and immunomodulatory effects of recombinant interferon-alpha 2b (IFN) in patients with recurrent or metastatic squamous cell carcinoma of the head and neck (SCCHN). Seventy-one patients with recurrent or metastatic SCCHN were entered into a phase II noncomparative randomized trial of IFN at two dosage schedules. Eligible patients with histologically proven SCCHN were randomized to receive low-dose IFN, 6 x 10(6) U/m2 daily x 3 every 4 weeks or high-dose IFN, 12 x 10(6) U/m2, 3 x/week. Pretreatment levels of natural killer (NK) activity, CD3, CD4, CD5, CD8, CD16, CD19, CD56, DR, and the CD4/CD8 ratio were evaluated for any relationship with survival. The toxicity encountered in patients receiving low-dose IFN was for the most part mild to moderate. With high-dose IFN, toxicity was greater with significantly more episodes of grade 3 and 4 toxicity encountered. Dosage reduction was required in the majority of patients receiving high-dose IFN. Of the four lethal complications, only one was thought to be possibly associated with therapy. Of the 32 evaluable patients receiving low-dose IFN, there were 1 complete response, 1 stable disease, 24 patients with progressive disease, and 6 unevaluable. Of the 29 evaluable patients taking high-dose IFN, there were 2 complete responses, 7 with stable disease, 16 with progressive disease, and 4 patients were unevaluable. Median survival in the two arms was similar (6.2 months). Because it was postulated that a more prolonged exposure to IFN might be needed for it to be effective, patients receiving > or = 6 weeks of therapy were evaluated. Median survival in that subset was 10 and 12 months for patients receiving low- and high-dose IFN, respectively. None of the immune parameters tested was a significant predictor of survival when evaluated in all cases entered into study regardless of therapy duration. No difference in baseline NK activity was noted between patients who received < 6 or > or = 6 weeks of IFN (p = 0.90). However, among the 35 patients who received > or = 6 weeks of therapy, a high baseline NK activity was a significant predictor of the duration of survival (p = 0.04). IFN was well tolerated in patients with recurrent or metastatic SCCHN. The higher incidence of toxicity encountered in the high-dose arm could be ameliorated by reducing the dose 50%. In patients receiving 6 or more weeks of therapy, elevated baseline NK activity was associated with increases in survival, suggesting that IFN may play an immunomodulatory role. Although the overall response rates were low, disease stabilization was noted, suggesting an antiproliferative, noncytotoxic role of IFN in this group of heavily pretreated patients.