Ninety-two patients with advanced bronchogenic carcinoma were prospectively randomized according to performance status, histology, and extent of disease to methyl-CCNU alone; methyl-CCNU and vincristine; or methyl-CCNU, vincristine, and methotrexate. Seventy-three patients were evaluable. Randomization to methyl-CCNU alone was discontinued when only three brief "static" responses were noted in 11 patients and the survival (p less than .01) and time on study (p less than .05) were noted to be significantly less than with the combination. Two of 32 patients treated with methyl-CCNU and vincristine, and two of 30 patients treated with methyl-CCNU, vincristine, and methotrexate had mixed responses with a median of 67.5 days. A static response was seen in eight of 32 and none of 30 patients, respectively. Minimal toxicity occurred in all regimens. Methyl-CCNU alone or in combination with vincristine or vincristine and methotrexate is of limited benefit in patients with lung cancer, although our data suggest (p = 0.15) that the addition of methotrexate increases time on study and survival in patients with extensive disease.