We conducted a prospective, randomized study to clarify the role of radiotherapy of the primary tumor in limited small-cell cancer of the lung. After stratification for sex and for performance score based on the ability to ambulate, patients were randomly assigned to receive initial radiotherapy plus chemotherapy, delayed radiotherapy plus chemotherapy, or chemotherapy alone. The chemotherapy consisted of cyclophosphamide, etoposide (VP-16-213), and vincristine, with doxorubicin subsequently replacing etoposide in alternate cycles 7 through 18. Chemotherapy was given every three weeks for 18 months. The radiotherapy comprised 4000 rad in four weeks, followed by a 1000-rad "boost" directed against residual disease. All patients received prophylactic whole-brain radiation. The patients enrolled totaled 426, and 399 were evaluable. There was a statistically significant difference in the frequency of complete responses in favor of the two radiotherapy regimens (P = 0.0013). Failure-free survival was also longer with these two regimens (P less than 0.001), as was the interval before treatment failure in the chest (P less than 0.001) and overall survival (P = 0.0099). As expected, toxic effects--chiefly neutropenia--were also increased. The addition of radiotherapy of the primary tumor to combination chemotherapy improved both complete-response rates and survival, with increased but acceptable toxicity.