It is increasingly appreciated that the pathogenic mechanisms of type 1 diabetes involve both the autoimmune aggressors and their beta cell targets, which engage in a conflicting dialogue within and possibly outside the pancreas. Indeed, autoimmune CD8 + T cells, which are the final mediators of beta cell destruction, circulate at similar frequencies in type 1 diabetic and healthy individuals. Hence a universal state of 'benign' islet autoimmunity exists, and we hypothesise that its progression to type 1 diabetes may at least partially rely on a higher vulnerability of beta cells, which play a key, active role in disease development and/or amplification. We posit that this autoimmune vulnerability is rooted in some features of beta cell biology: the stress imposed by the high rate of production of insulin and other granule proteins, their dense vascularisation and the secretion of their products directly into the bloodstream. Gene variants that may predispose individuals to this vulnerability have been identified, e.g. MDA5, TYK2, PTPN2. They interact with environmental cues, such as viral infections, that may drive this genetic potential towards exacerbated local inflammation and progressive beta cell loss. On top of this, beta cells set up compensatory responses, such as the unfolded protein response, that become deleterious in the long term. The relative contribution of immune and beta cell drivers may vary and phenotypic subtypes (endotypes) are likely to exist. This dual view argues for the use of circulating biomarkers of both autoimmunity and beta cell stress for disease staging, and for the implementation of both immunomodulatory and beta cell-protective therapeutic strategies.