An assay which could prospectively predict a response to chemotherapy prior to instituting therapy would be beneficial, especially for nonresponders. We evaluated at random a total of 50 patients with previously untreated stage III and IV head and neck squamous cell carcinoma for T-cell subsets and levels of circulating immune complexes (CIC). Of this group 23 patients had induction chemotherapy as their first modality of treatment. These patients received standard protocols of therapy consisting of cisplatin, 5-fluorouracil, methotrexate, and bleomycin. Of the laboratory measures assessed, only CIC correlated with clinical response to chemotherapy [no response (NR), partial response (PR), complete response (CR)] at the completion of the induction period (two to four cycles). Levels of CIC were determined by a polyethylene glycol (PEG) precipitation assay, and measured by spectrophotometry at 280 nm. In our population of all head and neck cancer patients, CIC were elevated (mean +/- SD values: 0.475 +/- 3.8 compared with controls: 0.184 +/- 0.07). In the NR group, the mean +/- SD was 0.707 +/- 0.43 (P = 0.001). Complete responders had a mean value of 0.332 +/- 0.21; partial responders had a mean value of 0.255 +/- 0.13. On the basis of the values determined, patients with markedly elevated levels of CIC would be predicted to respond poorly, or not all, to current induction chemotherapy protocols.