Are financial intermediaries inherently unstable? If so, why? What does this suggest about government intervention? To address these issues we analyze whether model economies with financial intermediation are particularly prone to multiple, cyclic, or stochastic equilibria. Four formalizations are considered: a dynamic version of Diamond-Dybvig banking incorporating reputational considerations; a model with delegated investment as in Diamond; one with bank liabilities serving as payment instruments similar to currency in Lagos-Wright; and one with Rubinstein-Wolinsky intermediaries in a decentralized asset market as in Du¢ e et al. In each case we find, for different reasons, financial intermediation engenders instability in a precise sense