This paper explores the effect of time-varying velocity on output responses to policies for reducing/stopping inflation. We study a dynamic general equilibrium model with sticky prices in which we introduce time-varying velocity. Specifically, we endogenize time-varying velocity into the model developed by Ireland (1997) for analyzing optimal disinflation. The nonlinear solution method reveals that, depending on velocity, the “disinflationary boom” found by Ball (1994) may disappear even under perfect credibility and that early output losses may be much larger than previously thought. Indeed, we find that a gradual disinflation from a low inflation may even be undesirable.