Internet software is an emerging software paradigm. Traditional evaluation methods for software reliability are no longer applicable because of the open and dynamic characteristics of the Internet software. In order to evaluate its reliability accurately, there must be a dynamic and open reliability evaluation approach. This paper presents the collection approach of the reliability-related data for the Internet software and its components. The paper also designs a software system for the reliability evaluation. The approach bases on the characteristics of Internet software; it uses aspect-oriented programming and pattern programming techniques to trace component real-time running data and save the data to a database. The approach can trace different granularity data according to the performance requirements of the system. The reliability evaluation system can predict the components and system reliability using the real-time data. This makes the components be selected expediently when the Internet software system is assembled dynamically. A case study is presented to illustrate the effectiveness of this approach