In a study of 411 patients with small-cell lung cancer (SCLC) entered on therapeutic clinical trials between 1973 and 1987, we analyzed whether changes in the prognostic importance of pretreatment factors had occurred during the 14-year time period. After adjusting for other prognostic factors, brain involvement was associated with shorter survival in patients treated before December 1979 (P = .024) but not in patients treated thereafter (P = .54). The patients diagnosed before 1979 had brain metastases documented by radionuclide scan while computed cranial tomography (CCT) was more commonly used after 1979. Patients who had brain metastases diagnosed by radionuclide scan lived a shorter period of time than patients who had the diagnosis made by the more sensitive CCT scan (P = .031). In contrast, Cox proportional hazards modeling showed that liver metastases in patients were associated with shorter survival in patients treated after 1979 (P = .0007) but not in patients treated before then (P = .30). A larger proportion of patients had a routine liver biopsy before 1979 than after 1979 when more patients had the liver staged with less sensitive imaging studies and biochemical parameters. Patients with SCLC whose cancer was confined to the thorax but had medical or anatomic contraindications to intensive chest radiotherapy had similar survival compared with patients with limited-stage SCLC who were treated with combination chemotherapy alone (P = .68). From these data we conclude: (1) the sensitivity of the staging procedures used can affect the impact on survival of cancer involvement of a given site; and (2) patients with cancer confined to their chest with medical or anatomic contraindications to chest radiotherapy do not have a shorter survival than patients with limited-stage disease treated with chemotherapy alone.