While a streaming-XPath engine provides a more efficient XPath evaluation performance than a DOM based engine, it still requires time to parse a whole XML document at least once. Taking into consideration that XPath is used to extract data from Web Service messages, the target element may be in the first part of documents, such as the SOAP Headers. Therefore, it is wasteful to parse a document all the way to the end. We present a new XPath engine named "XScope," which eliminates parsing XML documents by Schema information. In this paper, we describe the XScope approach, then evaluate its performance and show that the engine works effectively in Web Services.