This phase I/II study sought to determine the response rate and toxicity profile of escalating doses of paclitaxel (Taxol; Bristol-Myers Squibb Company, Princeton, NJ) when administered with fixed doses of cisplatin with granulocyte colony-stimulating factor support in 28 patients with head or neck cancer. The study was designed as a modified dose-finding trial and contained five dose-escalation levels of paclitaxel. Dose level 1 contained seven patients, and doses ranged from 175 to 220 mg/m2; dose level 2, 230 to 250 mg/m2 (five patients); dose level 3, 250 mg/m2 only (four patients); dose level 4, 260 to 280 mg/m2 (six patients); and dose level 5, 280 to 300 mg/m2 (six patients). Dose levels greater than 200 mg/m2 were supported with concomitant granulocyte colony-stimulating support. Paclitaxel was given on day 1 by 3-hour infusion; cisplatin 75 mg/m2 was given on day 2. Courses were given every 3 weeks. All patients were evaluable for toxicity and 27 were evaluable for response. The overall response rate was 77% (10 complete responses, 11 partial responses, four no change, and two disease progression). Over a median follow-up of 15 months (range, 7 to 22 months), 16 patients showed no evidence of disease and three are alive with disease. A dose-effect relationship was found between paclitaxel and peripheral neuropathy. Twenty-seven of 28 patients experienced alopecia, peripheral neuropathy, myalgias, and arthralgias. Most toxicities were grade 1 or 2; the mean duration of symptoms was 7 days. No dose-limiting hematologic toxicity was observed, nor was any significant neutropenia seen in those patients receiving filgrastim. The paclitaxel/cisplatin combination was found to be an effective first-line regimen for patients with head or neck cancer. Although the number of patients in this study was small, no relationship was noted between patient response and disease site.