There is a consistent pattern in the way that people predict painful experiences, natural or contrived, and this pattern resembles the way in which people predict frightening experiences. People tend to overpredict how much pain they will experience. These predictions of intensity tend to decrease after people have made an overprediction, and to increase if they underpredict the pain. After a correct prediction, the subsequent prediction tends to be unchanged. An underpredicted pain is experienced as being more aversive than a correctly or overpredicted pain. It is probable that people tend to overpredict the intensity of a variety of aversive experiences, and that underpredictions are followed by immediate and prolonged increases in the predictions of subsequent aversive events, unless there is a superordinate predictive pattern. The functional value of overpredictions of pain is considered, as are the clinical implications of these findings. © 1991.