The scalability of a feedback aggregating overlay is the most fundamental requirement for large-scale P2P computing. In Peer-To-Peer (P2P) trust management feedback provides an efficient and effective way to build a reputation-based trust relationship among peers. In this paper, the authors proposed a Scalable Feedback Aggregating (SFA) overlay for large-scale P2P trust evaluation. The local trust rating method is defined based on the time attenuation function, which can satisfy the two dynamic properties of trust. More importantly, based on the SFA overlay, an adaptive trustworthiness computing method can be defined the SFA-based trust model has greater adaptability and accuracy in handling various dynamic behaviors of peers.