This paper investigates the properties of optimal disinflation policies in a 'New-Keynesian' monetary general equilibrium model, seeking to identify conditions under which: (i) the optimal disinflationary path involves significant output losses; (ii) a 'rapid' disinflation is preferable to a 'gradual' one. According to our results, different degrees of monetary frictions and inertia in output or inflation may lead to different optimal disinflationary policies - some of which will be quick and painless, while others will proceed slowly and generate deep recessions.