In order to give a more refined notion of information that is conveyed by ambiguous expressions, this paper analyzes how ambiguity affects the knowledge of the dialog participants and, especially, what they know about each other after an ambiguous sentence has been uttered. Methods of processing ambiguity based on Kripke structures in a multi-agent system is presented. The information states of multi-agent system are represented as a Kripke structures. The agents communicate with each other by means of a tell function, whose application is constrained by an implementation of some of Grice´s maxims. The method can easily distinguish semantic ambiguity from perceived ambiguity in the information conveyed by ambiguous sentences, so it is of practical value for researching on multi-agent systems.