每個系統在佈建後能達到的服務容量一直是系統商最重視的問題。點對點(P2P)視訊串流系統的容量受到系統諸多動態變因影響，例如：每個節點的上下傳頻寬、點對點網路拓樸形成方法、影音延遲限制、影音串流位元傳輸率大小…等。在本篇論文中，我們研究樹狀拓樸的點對點視訊串流系統並且證明找到最佳化樹狀拓樸以達到最大系統容量是NP-hard問題；接著我們說明頻寬與視訊延遲將造成系統容量的限制；最後我們提出一個適合在動態環境下使用的點對點網路生成演算法，並且以模擬結果證明我們提出的演算法能同時考慮節點上傳頻寬與視訊延遲，進而達成較佳的頻寬使用率與系統容量。

People always want to know achievable service capacity while deploying new systems. The capacity ofP2P streaming systems changes due to dynamic system composition. Different formation parameters, likeuplink/downlink bandwidth of each peer and server, overlay network topology, video delay constraint, videostream bit-rate, etc, can result in different system capacity. In this paper, we focus on tree-based P2Pstreaming systems and deal with the routing and peer selection strategies. We prove that finding the routingtree to maximize system capacity is NP-hard. A distributed heuristic which considers both on peer uplinkbandwidth and video delay constraints is proposed. Simulation results show that the proposed heuristicachieves larger system capacity.