The article reviews the proposals and experience of the Pigou taxes (“bank levies”) that were implemented on financial institutions in the wake of the 2007-2008 financial crisis aiming at internalizing systemic risk or other externalities stemming from financial activities, in contrast to direct control of financial activities. It aims at explaining the reasons for the reluctance to generalize this instrument and to broaden its applications. First, introducing a Pigou tax requires several conditions that are not often met. Then, there are several operational bottlenecks when implementing a tax on financial activities in terms of calibration, management of asymmetric information and moral hazard, as well as regarding the governance of the tax. Finally, several instruments of direct control, like the separation of risky activities (e.g. separation of market activities by banks non directly participating in the financing of the economy), appear to be more efficient to ensure financial stability than a Pigou tax. However, the latter may be viewed as close to other instruments of direct control that also affect financial institutions' incentives, like the resolution of distressed financial institutions, but resolution authorities have been granted legal powers that go far beyond simple taxation.