An ad-hoc network is a set of limited range wireless nodes that function in a cooperative manner so as to increase the overall range of the network. Each node in the network pledges to help its neighbours by passing packets to and fro, in return of a similar assurance from them. All is well if all participating nodes uphold such an altruistic behaviour. However, this is not always the case and often nodes are subjected to a variety of attacks by other nodes. These attacks range from naive passive eavesdropping to vicious battery draining attacks. Routing protocols, data, battery power and bandwidth are the common targets of these attacks. In order to overcome such attacks a number of routing protocols have been devised that use cryptographic algorithms to secure the routing mechanism, which in turn protects the other likely targets. A limiting requirement regarding these protocols is the reliance on an omnipresent, and often omniscient, trust authority. In our opinion, this reliance on a central entity is against the very nature of ad-hoc networks, which are supposed to be improvised and spontaneous. We present in this paper, a trust-based model for communication in ad-hoc networks that is based on individual experience rather than on a third party advocating trust levels. The model introduces the notion of belief and provides a dynamic measure of reliability and trustworthiness in pure ad-hoc networks.