Web services have been widely employed in E-commerce system. It is a flexible way to use third-party applications for complex business logics implementation, which supports data transparent transmission via public interfaces. However, the single service with a fixed function has limitations to the diversity and complexity of user requirements, especially the interoperability of applications interactions. To address the challenge, the capability of dynamic services composition aims to select services and assign them to business process. However, there are kinds of possible composition plans to satisfy requested functions. It is a great importance to evaluate these plans to guarantee the solution of services composition is correctness and acceptable. In this paper, it proposes a service process verification approach to probabilistic spanning tree for dynamic composition. First, the probabilistic spanning tree is introduced to describe services composition plans according to the interface operations decomposition of given service requirements. Second, the service process model is generated from probabilistic spanning tree based on different pruning strategies, where the formal verification is performed to check each solution against quantitative properties related to the functional and non-functional requirements. Our method is feasibility to dynamic services composition since the quantitative verification guarantees the performance of composition solution.