The tumour microenvironment is a complex mix of cancer, stromal and reactive inflammatory cells. The association between inflammatory infiltrates and better cancer outcomes in the absence of immunotherapy likely reflects a measure of tumour antigen recognition and associated immune control. More recently inflammation, INFg and checkpoint ligand expression have been identified as response predictors, particularly for agents that block the PD1/PDL1 axis. Since PD1 inhibition is the emerging cornerstone of cancer immunotherapy, a variety of strategies are being evaluated to increase intratumoral inflammation with the intention of enhancing PD1/PDL1-inhibitor However this may be overly simplistic. Inflammation can vary in the localisation, intensity, cellular composition, regulatory mechanisms, antigen profile and the expression of inhibitory pathways. Equally, the absence of inflammation can reflect a variety of different mechanisms, ranging from low antigen burden resulting in a lack of intrinsic immunogenicity to the exclusion of inflammatory cells as a consequence of oncogenic signalling pathways; or might reflect induced regulatory processes. These different mechanisms point to the likelihood that the optimisation of therapy will require different therapeutic approaches. Rather than building empiric combinations, a diagnostic sub-classification of tumours may be required, based on the better understanding stromal biology and inflammatory processes. This would enable immunotherapy to be personalised, on the basis of biomarkers that reflect targetable biological mechanisms with the tumour.