Although insider selling is a potentially important proxy for undisclosed bad news, the literature has failed to document consistent evidence of insiders’ sales being informative. The lack of information in insider sales is attributed to researchers’ inability to separate liquidity-motivated from information-based insider trades. We present and evaluate a new algorithm for identifying information-based insider sales. We hypothesize that when individuals who have insider-status with respect to multiple firms sell shares of one firm in which they are insiders and at the same time buy shares of other affiliated firms, the sale is more likely to be information-based, since the proceeds are reinvested. Insider sales unaccompanied by insider purchases are more likely to be liquidity-motivated. We find that insider sales identified using this algorithm as information-based are followed by statistically and economically significant negative abnormal returns. In validation tests comparing the incidence of adverse events for firms with information-based versus liquidity-motivated sales in the year of and the two years following the sale, we find that information-based sales are significantly more likely to be associated with delistings, earnings declines, downward analyst forecast revisions, analysts dropping coverage, and earnings restatements. We conclude that it is possible to ex ante and directly identify insider sales with significant information content. Our results will be of interest to investors as well as to regulators designing insider trading rules.