Negotiation is a fundamental mechanism in distributed multi-agent systems. Since negotiation is a time-spending process, in many scenarios agents have to take into account the passage of time and to react to uncertain events. The possibility to decommit from a contract is considered a powerful technique to manage this aspect. This paper considers interactions among self-interested and autonomous agent, each with their own utility function, and focuses on incomplete information. We define a negotiation model based on asynchronous message passing in which the negotiation doesn't end when an agreement is reached but when the consequences of the contract have happened