This paper re-considers cointegrating behaviour between forward and spot exchange rates and the implications for the forward rate unbiasedness hypothesis. Extant empirical evidence examining forward and future spot rates is mixed, offering results both for and against cointegration; the forward rate as an unbiased predictor of the spot rate; and the existence of a time-varying risk premium. However, recent research has suggested that such analysis may be subject to bias and that models of cointegration between forward and current spot rates should instead serve as a starting point for analysis of exchange rate behaviour. Johansen cointegration analysis supports this contention showing that erroneous inferences can be made by merely using future spot rate data. Subsequently both single equation and panel estimation methods support cointegration between forward and current spot rates, but that the forward rate is a biased predictor. Further, single equation tests are conducted over a rolling window of five years through our sample. These results largely confirm those for the full sample.