In patients with brain metastases from non-small cell lung cancer, the prognostic impact of primary tumour histology, a feature with increasing implications for choice of systemic therapy, is not well defined. Therefore, a multi-institutional analysis was performed: retrospective uni- and multivariate analyses in 209 patients treated with different approaches including surgery and radiosurgery. While squamous cell and large cell carcinoma patients had comparable survival, those with adenocarcinoma survived significantly longer. In multivariate models, adenocarcinoma histology was confirmed as independent prognostic factor, which complements both recursive partitioning analysis (RPA) classes and diagnosis-specific graded prognostic assessment (GPA). When evaluated together with primary tumour control, extracranial metastases, number of brain metastases, age and performance status as individual covariates rather than RPA or GPA score, adenocarcinoma histology again emerged as significant prognostic factor. A significant but small survival advantage for patients with adenocarcinoma was evident already in the time period before drugs such as pemetrexed and epidermal growth factor receptor tyrosine kinase inhibitors were available. However, the gap has widened in recently treated patients. Comparable to patients without brain metastases, primary tumour histology should be taken into account when assessing patients' prognosis and recommending treatment strategy.