Rare cancers, which collectively account for almost 25 % of all malignancies, are poorly understood in terms of their aetiology and pathogenesis and are infrequently the focus of translational and clinical research to improve their diagnosis and treatment. Consequently, those affected have comparatively few treatment options, and their prognosis is worse than that of patients with more common entities. Here we review two relevant groups of rare cancers, bone and soft-tissue sarcomas and neuroendocrine tumours (NET), to illustrate recent efforts towards individualised, biology-guided clinical management to improve long-term outcomes. Specifically, we address how comprehensive, multi-layered molecular analyses, including the assessment of predisposing hereditary factors, and innovative imaging approaches can improve the diagnosis of these diseases, allow for better prognostic assessment, and provide new targets for pharmacologic and, in the case of NET, nuclear medicine interventions, whose clinical value must be determined in controlled trials optimally tailored to the particular patient population most likely to benefit. Furthermore, we describe the importance of multidisciplinary collaboration in dedicated reference centres for rare cancers and the increasingly acknowledged potential of networking across institutions at a national and international level. Finally, we illustrate the value of a learning health system based on the systematic collection and sharing of the biological and clinical profiles of patients with rare cancers to achieve continuous cross-fertilisation of scientific and clinical efforts, making the vision of stratified precision medicine in these long-overlooked diseases a reality.