In this paper we assess the rationality and goodness-of-fit of macroeconomic forecasts for 21 large emerging market countries during the past two decades. We find that in some countries forecasts have not been rational in the sense that they seem to have a systematic bias. We also find that for both GDP growth and inflation the forecast errors are larger for more volatile economies. For some countries GDP and inflation forecast errors have positive correlation, which is consistent with New Keynesian macromodels, whereas for other countries this is not the case. Our results suggest that relying on any one forecast methodology is not prudent.