Bail-in is hailed as the resolution tool that could end the ‘too-big-to-fail’ (TBTF) problem. Bail-in allows to absorb losses and recapitalise a bank by writing down and/or converting to equity its debt. In spite of the high hopes vested on bail- in, will resolution via bail-in always be feasible without spurring contagion and financial instability? To bail-in or not to bail-in. That is the question. What is the answer? This question is still open for debate. Some eminent academics, such as Charles Goodhart (see e.g. Avgouleas and Goodhart (2015)), are skeptical that bail-in would work in system-wide financial crises. Using a novel system- wide stress testing approach based on Farmer et al. (2019), we show that bail- in usually works in idiosyncratic cases of bank failure. However, we find that bailing in banks in systemic crises may exacerbate financial fragility if bail-in policy parameters are not set appropriately. In such case (the risk of) bail-ins can set in motion a concatenation of contagion effects that include exposure losses, re- valuations of bail-inable debt, runs on bail-inable debt and firesales. Strikingly, if resolution authorities do choose the bail-in policy parameters appropriately we observe a pronounced phase shift from an unstable to a stable financial system. Stability remains even if systemic effects require multiple banks to be bailed-in. Our evidence suggests that the pivot for stability is fortunately in hands of policymakers. But that the current policy settings are in the regime of instability.