The authors treated 32 patients with Stage IIIB or IV non-small cell lung cancer (NSCLC) with an outpatient regimen of edatrexate (10-ethyl-10-deaza-aminopterin) (10-EdAM) on days 1 and 8, cyclophosphamide on day 1, and cisplatin on day 1, repeated every 3 weeks with dose modification. The 22 men and 10 women (median age, 57 years of age) had no prior chemotherapy and a Zubrod performance status less than or equal to 2. A schedule with initial doses of 80 mg/m2, 800 mg/m2, and 80 mg/m2, respectively, yielded a 47% major response rate with two complete responses (95% confidence interval [CI], 25% to 70%), but it also yielded significant stomatitis and myelosuppression. A schedule with reduced starting doses (70 mg/m2, 700 mg/m2, and 70 mg/m2) was better tolerated, but dropped the major response rate to 27% with no complete responses (95% CI, 11% to 52%). Median survival time was 39 weeks for all 30 evaluable patients without a significant difference between the treatment groups (which were comparable in patient characteristics). Major response, however, was associated with longer survival time than minor response or no change (P = 0.024) or progressive disease (P = 0.001) (median survival times, 55, 39, and 27 weeks, respectively). When the doses delivered were compared, patients treated with the reduced dose schedule received less mean 10-EdAM per course (P = 0.01), although the doses of cyclophosphamide and cisplatin were comparable to the original dose schedule for the second course and thereafter. These results suggest that this three-drug regimen may have synergistic antitumor effects, with a steep dose-response relationship, particularly with 10-EdAM. With amelioration of the dose-limiting stomatitis of 10-EdAM, it seems possible to maximize the antitumor effects of this regimen.