The issue whether cranial radiotherapy (RT) should be used prophylactically (PCI) or therapeutically (TCI) in small cell lung carcinoma (SCLC) is considered controversial by some oncologists. Trying to clarify this issue we have performed a retrospective analysis of a Southwest Oncology Group (SWOG) protocol for disseminated SCLC. Three Hundred and seventy-seven cases had no evidence of metastases to the brain (MB). One hundred and forty four of those had PCI. Seventy one cases were diagnosed of MB, and 64 received TCI. We confirmed previous reports showing a low percentage of brain relapse with PCI (around 5%), with minimal immediate morbidity. We also confirmed a high percentage of objective response (90%) with TCI, (although we had no response information in 40% of them) with long duration of response of 33 weeks. Brain relapse after TCI was only 18%. Only long-term survivors had brain relapse as survival of relapsing patients was longer than those without brain relapse (45 weeks versus 33 weeks, p = 0.06). However, 20 (31%) of the 65 with initial MB died within 6 weeks of registration, some without completing RT to brain. In the majority, cause of death was considered related directly to brain damage, or indirectly as sepsis developed in patients whose poor performance status was considered to be caused by their brain symptoms. When comparing patients with and without MB, the former had (a) worse survival (24 versus 32 weeks, p = 0.02) and (b) higher proportion of patients with poor initial performance status (50% versus 34%, p = 0.04). Although the possibility of long-term morbidity with PCI is deterring some oncologists from recommending it, our data show that MB creates a real chance for immediate morbidity and this should not be ignored. The pros and cons of both approaches and some new recommendations for PCI are discussed.