As businesses begin to link Web services to create new functionality in the form of composite Web services, known as Web processes, it will become increasingly important to have a way of measuring their quality of service (QoS). To this end, we present a method that uses a predictive QoS model to compute the QoS for Web processes in terms of performance, cost and reliability. The ability to compute QoS for a Web process enables an organization to tune the process. Tuning Web processes presents an interesting problem. During the act of tuning, a business may want to explore many different configurations of the Web process in order to answer “what-if” questions. Composing and evaluating the QoS for many different configurations may be prohibitive in terms of time and costs. We present a simulation based technique to overcome this challenge to tuning Web processes.