Since web services intended for the same application environment tend to be functionally homogeneous, researchers have turned to their non-functional aspects in order to constructively compare web services and choose an appropriate one. In general, quality of service is very important to users. Many schemes that consider quality assessment have been proposed for web service selection. With the increasing number of qualityrelated attributes, an effective quality assessment method needs satisfactory scalability. Schemes based on the analytic hierarchy process(AHP) meet this requirement. However, prevalent methods in this vein overlook the fact that the traditional AHP needs a complete judgment matrix. In practice, all the information needed to construct a judgment matrix is often unavailable for a variety of reasons, due to which certain judgments cannot be made. In this paper, for an incomplete judgment matrix, we propose an improved AHP approach to consistency verification and the subsequent ordering. Our method can deal with situations where information is insufficient,and inherits all the merits of the traditional AHP approach. A case study establishes the effectiveness of our proposed method. Since web services intended for the same application environment tend to be functionally homogeneous, researchers have turned to their non-functional aspects in order to constructively compare web services and choose an appropriate one. Many schemes that consider quality assessment have been proposed for web service selection. With the increasing number of qualityrelated attributes, an effective quality assessment method needs vigilance. Schemes based on the analytic hierarchy process (AHP) meet this requirement. However, prevalent methods in this vein overlook the fact that the traditional AHP needs a complete judgment matrix. In practice, all the information needed to construct a judgment matrix is ​​often unavailable for a variety of reasons, due to which certain judgments can not be made. In this paper, for an incomplete judgment matrix, we propose an improved AHP approach to consistency verification and the subs equent ordering. Our method can deal with situations where information is insufficient, and inherits all the merits of the traditional AHP approach. A case study establishes the effectiveness of our proposed method.