The goal of this National Cancer Institute-sponsored phase I trial is to determine the feasibility, toxicity, and pharmacokinetics of continuous-infusion (24 hr/d, 7 d/wk, 7-week total) intravenous paclitaxel (Taxol; Bristol-Myers Squibb Company, Princeton, NJ) combined with standard curative radiotherapy (RT) for previously untreated, locally advanced head and neck squamous cell cancers. Eligible patients have squamous cell cancers of the head and neck with expected 5-year survival rates of < or =25%; a good performance status; adequate hematologic, hepatic, and renal functions; and no distant metastases. All patients receive 70 Gy megavoltage RT in 7 weeks (2 Gy/d x 5 d/wk). Paclitaxel is delivered by protracted venous infusion starting 48 hours before RT and continuing for its duration. Biopsies for cell-cycle distribution analyses and paclitaxel tissue levels are obtained, if possible, before beginning paclitaxel and after 48 hours just before RT begins. The dose of paclitaxel is escalated in cohorts of three patients. Eighteen patients are evaluable for toxicity. Treatment has been completed through the 6.5 mg/m2/d dose level and is ongoing at 10.5 mg/m2/d. There has been no dose-limiting toxicity thus far. With the exception of anemia, toxicity is commensurate with what would be expected from RT alone. A slowly progressive normocytic anemia with no renal dysfunction was found to be associated with an acquired hypoerythropoietin state. Tumor biopsies have suggested the possibility of paclitaxel-induced mitotic arrest. This therapy is feasible and has been well tolerated through current dose levels with no dose-limiting toxicity. There is a suggestion of biologic activity evidenced by the anemia and the possibility of alteration in cell-cycle distributions. Dose escalation is ongoing.