We used the yen/dollar exchange-rate forecasts of the Wall Street Journal (WSJ) poll to analyse whether exchange-rate forecasters have an asymmetric loss function. To this end, we applied an approach recently developed by Elliott et al. (2005). We found that only few forecasters seem to form forecasts under an asymmetric loss function. For some forecasters, accounting for the asymmetry of their loss function makes their forecasts look rational.