This paper studies a New Keynesian model in which monetary policy may switch between regimes. We derive sufficient conditions for indeterminacy that are easy to implement and we show that the necessary and sufficient condition for determinacy, provided by Davig and Leeper, is necessary but not sufficient. More importantly, we use a two-regime model to show that indeterminacy in a passive regime may spill over to an active regime no matter how active the latter regime is. As a result, a passive monetary policy is more damaging than has been previously thought. Our results imply that the propagation of shocks in an active regime, such as that of the Federal Reserve in the post-1982 period, may be substantially affected by the possibility of a return to a passive regime of the kind that was followed in the 1960s and 1970s.