We examine the effectiveness of forward guidance at the effective lower bound (ELB) in the context of the COVID-19 pandemic. Survey evidence underscores the myopia of professional forecasters at the initial stages of the pandemic and the extraordinary dispersion of their recent forecasts. Moreover, financial markets are now practically certain that U.S. short-term nominal interest rates will remain at the ELB for the next several years; consequently, forward guidance would have to refer to a much longer time horizon than in previous experience. To analyze the effects of these issues, we consider a canonical New-Keynesian model with three modifications: (1) expectations formation incorporates the mechanisms that have been proposed for addressing the forward guidance puzzle; (2) the central bank has imperfect credibility in making longer-horizon commitments regarding the path of monetary policy; and (3) the central bank may not have full knowledge of the true structure of the economy. In this framework, providing substantial near-term monetary stimulus hinges on making promises of relatively extreme overshooting of output and inflation in subsequent years, and hence forward guidance has only tenuous net benefits and may even be counterproductive.