Using a closed-economy model, Jensen (2002) and Walsh (2003) have, respectively, shown that a policy regime that optimally targets nominal income growth (NIT) or the change in the output gap (SLT) outperforms a regime that targets inflation, because NIT and SLT induce more inertia in the actions of the central bank, effectively replicating the outcome obtained under precommitment. The author obtains a very different result when the analysis is extended to open-economy models. Flexible CPI-inflation targeting outperforms both SLT and NIT and is the most robust targeting regime. The gains from targeting CPI inflation are particularly large when the model features transmission lags and/or departures from the uncovered interest parity condition. The author also finds that the stabilization bias inherent in discretionary policy is smaller in an open-economy setting.