What drives leaders’ decisions about whether to continue or end an ongoing war? The private information explanation for war holds that leaders fight because they believe that doing so will advance national interests, and they settle hostilities when new information reduces their optimism about the possibility of long-term success. Yet significant theoretical disagreement exists about both the extent to which and the manner in which new information, especially battlefield information, promotes settlement. This article unpacks the logic of the informational mechanism, arguing that settlement will be more likely when there has been more extensive fighting and that countries are more likely to make concessions to end wars when battlefield results have deteriorated; short-term spikes in war intensity by contrast do not promote settlement. Moreover, building on work on leadership turnover and settlement, I show that leader replacement is sometimes part of the information-updating process, especially in autocracies: new leaders without political ties to the person in power at the start of the war are more likely both to come to power when war is going poorly and to end wars once in office. Tests of these arguments make use of new participant-level data on the timing of battle deaths for all Correlates of War interstate wars, which allows me to examine the effects of changing battlefield developments across a wide range of cases in a manner that was previously impossible.