I examine the Hansen and Sargent [2003. Robust control of forward-looking models. Journal of Monetary Economics 50, 581–604] formulation of the robust Stackelberg problem and show that their method of constructing the approximating equilibrium is generally invalid. I then turn to the Hansen and Sargent [2007. Robustness, manuscript (version dated March 22, 2007)] treatment, which, responding to the problems raised in this paper, changes subtly, but importantly, how the robust Stackelberg problem is formulated. In the context of Hansen and Sargent [2007. Robustness, manuscript (version dated March 22, 2007)], I prove, first, that their method for obtaining the approximating equilibrium is now equivalent to the one developed in this paper, and, second, that the worst-case specification errors are not subject to a time-consistency problem. In the context of the Erceg et al. [2000. Optimal monetary policy with staggered wage and price contracts. Journal of Monetary Economics 46, 281–313] sticky wage/sticky price model, I find that a robust central bank will fear primarily that the supply side of its approximating model is misspecified and that robustness affects importantly central bank promises about future policy.