This paper uses forecast data from 1995 through 2014 to examine, whether the market consensus of exchange rate forecasts has an effect on the forecasts of individual experts. Such an effect could take the form of herding or anti-herding. We use a very comprehensive data set to study experts' forecasts of three of the most important exchange rates. The results indicate that anti-herding vis-à-vis the consensus of forecasts occurs more often than herding. We also show how the increase in the forecasting horizon and financial crises affect the intensity of anti-herding behavior. Moreover, we report that the (anti-)herding behavior does not affect the forcasting performnce. •Foreign exchange forecasters strategically differentiate from each other.•Especially when issuing short-term forecasts, forecasters anti-herd rather than herd.•Anti-herding behavior does not seem to affect overall forecaster performance.•Financial crises only slightly decrease anti-herding behavior of forecasters.•Central banks should consider the behavior of forecasters in their forward guidance.