With the growing popularity and capabilities of mobile devices, peer-to-peer networking among such devices is increasingly of interest for mobile content sharing. One of the major challenges in practical use of Mobile Peer-to-Peer networks (MP2P) is the trust among peers. Traditionally, solutions in the state of the art have focused on a peer's past experience in evaluating trust of other peers, based on direct interactions. Previously unknown peers (with no history of direct interactions) are assessed based on third party recommendations, yet again requiring a peer to evaluate and find trustworthy recommenders. This reveals the fundamental need to find peers with honest intentions before any interaction. It becomes challenging when no known peers are in the vicinity, which is highly likely in an MP2P scenario. For a general mobile user, the probability of encountering trustworthy peers in particular situations or environmental contexts may be higher than in other contexts, e.g. in office than on the road while traveling. Further, observed peers which are co-located over a number of environmental contexts may have more in common and thus resulting a higher mutual trust. These facts can be utilized to enrich the trust derivation process in a decentralized manner. In this paper, we propose a generalized and a novel distributed mechanism to estimate the trust for peers using their encounter history in different environmental contexts, and a way to prioritize contexts depending on the level of association with them. When evaluated against real user data of the reality mining dataset, the results of the proposed mechanism show a significantly improved accuracy of trust evaluation compared to the state of the art.