This research applies the Einhorn-Hogarth ambiguity model to a two-party negotiation situation involving medical liability cases. In the first study, subjects who were assigned to the role of a defendant or a plaintiff in a medical liability case altered their out-of-court settlement offers according to the estimated likelihood of winning the case at trial, the amount of ambiguity associated with that estimate, and the importance of the case (p <. 05). As predicted by the Einhorn-Hogarth model, the effect of ambiguity on defendant subjects facing a potential loss differed (p <. 05) from its effect on plaintiff subjects facing a potential gain. The second study used the model to show that increasing the amount of ambiguity in negotiation situations can make the potential for a settlement less sensitive to parties having substantially different perspectives of their chances of winning at trial. These predictions were supported in an actual negotiation context. When each party had a very different estimate of the plaintiff winning the trial (p =. 20 vs p. 80), there were two to five times more settlements when there was much ambiguity than when there was little ambiguity (p <. 001). These results provide further support for the Emhorn-Hogarth ambiguity model. They also show that contrary to popular belief, increasing ambiguity in a negotiation context can increase the likelihood of a negotiated settlement. © 1993 Academic Press, Inc.