We study the validity of uncovered interest-rate parity by constructing ultra-long time series that span two centuries. The forward-premium regressions yield positive slope estimates over the whole sample. The estimates become negative only when the sample is dominated by the period of 1980s. We also find that large interest-rate differentials have significantly stronger forecasting powers for currency movements than small interest-rate differentials. Furthermore, when we regress domestic currency returns on foreign bonds against returns on domestic bonds as an alternative test of the parity condition, the null hypotheses of zero intercept and unit slope cannot be rejected in most cases. These results are consistent with a world in which expectations formation is highly imperfect and characterized on the one hand by slow adjustment of expectations to actual regime changes and on the other by anticipations for extended periods of regime changes or other big events that never materialize. An historical account of expected and realized regime changes adds credence to this explanation and illustrates how uncovered interest-rate parity holds over the very long haul but nevertheless can be deviated from over long periods of time due to ex post-expectation errors.