Service-oriented architecture has been widely applied to construct new software system or integrate some legacy applications. Web service technology is the most representative approach to build such system. However, this new software development manner inevitably brings plague to the latter maintenance activities such as comprehension and measurement, despite its convenience for constructing software systems. In the paper, the control flow in Web service composition is modeled by service control flow graph (i.e. SCFG). Then, the execution path complexity metric is proposed through introducing the knowledge about cognitive informatics. Based on such concept, the metric of control flow complexity for whole Web service-based system is also addressed. On the other hand, two data interaction complexities are also presented to reflect the information volume of message communications between services through analyzing the SOAP message. In addition, a well-known Web service composition is used as a case to demonstrate the above metrics. The analysis results show that our measurement methods are reasonable and effective.