Forty-seven patients with recurrent or metastatic head and neck squamous cancer were entered into a prospectively controlled study comparing high-dose infusion methotrexate with leucovorin against standard-dose methotrexate without leucovorin. Patients were stratified for prior treatment and hematogenous metastases prior to randomization. Patients received either methotrexate (1,500 mg/m2) infused over 24 hours with leucovorin or 40 mg/m2 intramuscularly. Each treatment was given weekly for the first six weeks and followed a dose escalation schedule to toxicity. After six weeks patients received the high-dose regimen every two weeks and the low-dose schedule continued weekly. One patient was ineligible and four had inadequate follow-up to assess response (less than two weeks). Forty-two patients were evaluable for survival and 37 for response. Six of 19 patients (32%) treated with high-dose infusion responded (one complete response) and four of 18 patients (22%) receiving standard dosage responded (P = .52). Treatment with high-dose methotrexate resulted in an improved duration of disease control over standard dosage with the respective median times to progression of 11 weeks and five weeks (P = .04). These differences were most marked in good performance status patients (P = .007) and those without hematogenous dissemination (P = .02). Toxicity, however, was also significantly worse in the high-dose treatment group (P = .01) and survival was identical between treatments (4.2 months). The authors conclude that any treatment advantage to high-dose methotrexate may be attributable to its greater toxicity rather than to a selective therapeutic effect and this regimen does not result in improved patient survival. Good performance status patients may benefit from more aggressive therapy.