Content-based publish/subscribe (CBPS) paradigm is a powerful data dissemination paradigm that offers both scalability and flexibility. However, its nature of high expressiveness makes it difficult to analyze or predict the behavior of the system such as event delivery probability and end-to-end delivery delay, especially when deployed over unreliable, best-effort public networks. This paper proposes an analytical model that abstracts both expressiveness of content-based publish/subscribe systems, and uncertainty of underlying networks. The overall goal of this model is to predict quality of service in terms of delivery probability and timeliness based on partial, imprecise statistical attributes of each component in the distributed CBPS system. The evaluation results via extensive simulations with real-world traces yield effectiveness of the proposed prediction model. The proposed prediction model can be used as a building block for automatic quality of service control in publish/subscribe systems such as subscriber admission control, broker capacity planning, overload management, and resource adaptation.