By incorporating new information generated by currency derivatives trading, underlying exchange rates should be less forecastable than previously and the underlying currency markets should, therefore, be more efficient. This hypothesis was tested, for the first time, for the period 1982 through 1997 on a clean sample of three major types of currency derivatives launched in two prominent markets. Various statistical tests indicate that following the introduction of the derivative contracts, the underlying exchange rates became more random and the currencies involved tended thus to be priced more efficiently, which supports the hypothesis.