Tumour classification is a pivotal process for the clinical management of cancer patients. It determines the diagnosis discussed with the patient and subsequently informs about surgical, radio-therapeutic and chemotherapeutic management. However, since many drug decisions are now based on a molecular target independent of the tissue of origin, there has been debate in the literature as to whether conventional histological classification remains useful. This is particularly notable in lung carcinoma where selection of therapy is determined by the molecular status of the tumour i.e. EGFR mutation, ALK rearrangement etc. Each of these sub-categories account for a small proportion of total number of lung cancers and in some subtypes could be construed as rare. This new paradigm of targeting gene mutations within cancer rather than necessarily treating the particular tumour type, is not only of great academic interest but has profound and long-lasting consequences on the practise of cancer medicine, on how pathology is configured, and on how clinical trials are designed in the multi-billion dollar route to drug approval. Indeed, it has been argued that the current method of tissue-based classification will actually hamper further incremental increases in survival and improved quality of life that has been seen over the last decades. Thus, patients should be classified by virtue of their molecular phenotype, regardless of tissue type, the corollary being all cancers are individual. This approach has the significant benefit of uncovering resistance mechanisms, as occurred with BRAF inhibitors that are effective in melanoma but not colorectal carcinoma. Such a strategy could be implemented immediately for rare cancers that have a significantly higher mortality compared with other tumour types.