As Peer-to-Peer (P2P) e-commerce environments lack central management, prior to new transactions with an unknown peer, the trust evaluation becomes a very important concern, which relies on the transaction history data. Traditionally the evaluation process is based on other peers? recommendations neglecting transaction amounts. This may lead to the bias to the transaction trust evaluation and risk the new transaction, which may occur between unknown peers. This paper presents a novel model for transaction trust evaluation, which distinguishes transaction amounts and thus computes different impact factors when analyzing old transactions and computing trust values. As a result, the trust evaluation becomes more accurate, which is dependant on transaction history, the amounts of old transactions, the amount of the new transaction, and the temporal dimension. Therefore the obtained trust value can be taken as the risk indication of the forthcoming transaction.