ABSTRACT The presence of brain metastases (BM) remains a therapeutic challenge as most trials with the exception of local treatment with surgery and radiosurgery in patients with solitary BM have failed. The negative trials suggest that the disease in the brain is not the principal determinant of life expectancy in patients with disseminated disease. Large trials without enrichment of the study population by a predictive biomarker are also unlikely to be successful. The blanket use of whole brain radiotherapy for multiple BMs is also no longer tenable. The concept of blood brain barrier, unlikely to be of significance in patients with enhancing brain metastases, should not be a bar to the use of chemotherapy, where the principal determinant of efficacy is likely to be chemoresponsiveness of systemic disease. Successful use of prophylactic brain treatment is predicated on the assumption that brain dissemination is a late metastatic event. Metastases suppressors, with mode of action spanning the metastatic cascade, have prophylactic efficacy in vivo, although the studies are mostly based on non-spontaneous models, which are unlikely to reflect the true clinical situation. Nevertheless some agents are approaching the time of clinical application. The successful development of new therapies for BMs needs refinement of the current trial design. Phase I/II studies of new agents should not exclude patients with BM to provide “proof of principle” of therapeutic efficacy in the brain. Subsequent trials, with survival as the primary endpoint, should focus on patient population where BM are likely to be the determinants of outcome (i.e. with inactive systemic disease) and enriched for a predictive biomarker. Large trials, even in a single tumour type, testing a single agent or a combination, either alone or in addition to standard treatment without appropriate patient selection are likely to continue to fail. Disclosure The author has declared no conflicts of interest.