This paper evaluates rigorously the predictive power of the head-and-shoulders pattern as applied to daily exchange rates. Though such visual, nonlinear chart patterns are applied frequently by technical analysts, our paper is one of the first to evaluate the predictive power of such patterns. We apply a trading rule based on the head-and-shoulders pattern to daily exchange rates of major currencies versus the dollar during the floating rate period (from March 1973 to June 1994). We identify head-and-shoulders patterns using an objective, computer-implemented algorithm based on criteria in published technical analysis manuals. The resulting profits, replicable in real-time, are then compared with the distribution of profits for 10,000 simulated series generated with the bootstrap technique under the null hypothesis of a random walk.