Two hundred thirty-nine patients with microscopically proven, inoperable bronchogenic carcinoma were allocated at random to receive twice weekly I.M. injections of either methotrexate at "high dose" of 0.06 mg/kg/dose or methotrexate at "low dose" of 0.2 mg/kg or visually indistinguishable placebo in the same volume of 0.1 ml/kg for four months. Twelve patients were invalidated for procedural reasons. Objective response (greater than or equal to 50% tumor regression) was dose-related with 21% of 48 patients with measurable disease on high -ose, 11% of 37 patients on low dose, and 6% of 32 patients on placebo. Corresponding response rates for epidermoid carcinoma were 35% of 23 patients, 9% of 11 patients, and 0 of 13 patients. Responders in the two treatment groups had a three to four fold increase of median survival (p less than .05). Non-responders on high and low dose methotrexate lived as long as patients on placebo. Leukopenic patients in all three treatment groups lived substantially longer than patients without leukopenia less than 4,500/mm3, irrespective of presence or absence of objective response. All three regimens were well tolerated. None of the patients had life-threatening toxicity. It is concluded that methotrexate at "high dose" is a potentially useful drug for temporary palliation of epidermoid carcinoma of the lung.