Short sellers are known to have private information about security prices. Empirical evidence of short selling, however, is based on only half of short sellers’ trading activity; specifically, the opening of the position. Using disclosed large-short-position data from the Japanese stock market, we provide the first detailed evidence of covering trades and find a positive reaction to short covering that only partially reverses. Although these results are consistent with substantial transaction costs for closing large short positions, they also reveal that some short sellers are privately informed about positive future events and have timing ability in covering positions.