This paper presents a specification-based robust testing framework for Web services. Web services testing is done by 1) extracting condition and event combinations from the Web service specification, 2) ensuring that these combinations are consistent with each other; 3) performing completeness analysis to identify all the missing condition and event combinations; 4) Identifying the locations where updates are needed to maintain the completeness and consistency.; 5) emphasizing on robustness testing by generating positive as well as negative test cases. An efficient algorithm called Covering Scenario Generation is proposed to identify the locations where incompleteness and inconsistency exist. The algorithm is based on the min-terms of Boolean expressions that combine multiple conditions into a single checkable item. The proposed algorithm has been experimented with several large industrial applications and the results indicated that the proposed algorithm is robust and scalable to large applications. A case study is designed to illustrate the design and testing process.