A total of 497 patients with histologically or cytologically confirmed small-cell lung cancer were prescribed initial treatment with six courses of etoposide, cyclophosphamide, methotrexate and vincristine at 3-week intervals. Patients with limited disease (74% of the total) also received radiotherapy (40 Gy in 15 fractions in 3 weeks) to the primary site between courses 2 and 3. At the end of this initial treatment, 265 patients still in complete or partial response were randomly allocated to six further courses of maintenance chemotherapy (M series: 131 patients) or to no maintenance chemotherapy (NoM series: 134 patients). Response, as assessed 3 weeks after the second course of initial chemotherapy, was achieved in 85% of the 264 patients assessed, a complete response in 11%. The median survival period from the date of start of chemotherapy was 39 weeks; 154 (31%) of the patients were alive at 1 year, 29 (6%) at 2 years and 17 (3%) at 3 years. The patients' general condition and extent of disease pretreatment correlated significantly with survival. Among the 131 M and 134 NoM patients there was no overall survival advantage to either series (P = 0.27, log rank test), although in 99 patients who had a complete response to initial chemotherapy as assessed at the time of randomisation there was a suggestion that survival was longer in the M series (P less than 0.05, log rank test), the median survival periods from the date of randomisation being 42 weeks for the M and 30 weeks for the NoM patients. Maintenance chemotherapy was associated with additional toxicity and a poorer quality of life as assessed intermittently by clinicians and daily by patients. In conclusion, no worthwhile clinical advantage was achieved by the policy of continuing chemotherapy beyond six courses, except possibly in patients with a complete response to the initial six courses.