Previous research has shown that presuppositions can alter memory, but these results depend upon a restricted class of pragmatic conditions. Given a specified source with a presumed intention to mislead, listeners might not enter such presuppositions into memory. In two experiments, subjects first observed an accident depicted in a series of slides. In Experiment 1, the leading questions with no source led subjects to “remember” the presupposed facts; attribution of the questions to a lawyer representing the defendant eliminated that effect. In the second experiment, the presuppositions were introduced in a transcript of an eyewitness account of the accident. If the account was from a neutral bystander, subjects “remembered” the presupposed facts; yet if the account came from the driver causing the accident, the same presupposed facts were not remembered. These results reflect the influence of pragmatic conditions on normal language processing, conditions normally excluded from laboratory experiments.