Reputation-based trust evaluation is critical to e-commerce or e-service applications. In some applications (such as eBay), the trust management mechanisms have been introduced to provide valuable information to buyers prior to placing orders and making payments. Meanwhile, the trust issue is also actively studied in the research community. However, in most existing studies, a single trust value is computed based on ratings given for a seller or a service provider to indicate the current trust level. This is interesting but too simple to reflect the service quality history and trust features well under some circumstances. It may also be misleading for the decision making of buyers or service customers. In this paper, we present a novel fuzzy regression based trust vector approach to depict the trust level with more indications, and predict the trustworthiness of a forthcoming transaction with its advertised QoS values and transaction history.