Between October 1981 and November 1984, 291 patients with inoperable advanced non-small cell carcinoma of the lung (NSCLC) were randomized to a two-arm study. Eighteen of 291 defaulted treatment and were excluded from the study. Twenty-seven of 273 died during treatment; they were invaluable for treatment response but were included in survival analysis. Without correction for lung attenuation 45 Gy/18 fractions/4 1/2 weeks were given in arm 1 and 31.2 Gy/4 fractions/4 weeks were given in arm 2. One hundred twenty-eight of 273 were included in arm 1 and 145/273 in arm 2. The two arms were comparable in patient age, sex, performance status and symptoms, primary tumor site, histology, stage of the disease, and distribution of metastases and radiation portal size used. Prognosis was poor with an overall median survival of 20 weeks and was similar in both arms. Radiological tumor response was also similar: 53% in arm 1 and 50% in arm 2. However arm 1 was superior than arm 2 in achieving symptom palliation, 71% vs 54%, p less than 0.02. Treatment complications were mild and included mainly radiation oesophagitis and pneumonitis and pulmonary fibrosis. Treatments in both arms were equally well tolerated.