Abstract: With the expansion of business intelligence (BI) market, usability evaluation of on-line analytical processing (OLAP) systems has become a hot issue in database industries. How to measure the performance of OLAP systems is an important efficiency aspect that needs to be resolved with an OLAP performance benchmark. This paper presents a unified testing approach for efficiency evaluation of both ROLAP and MOLAP implementation systems based on the APB-1 benchmark introduced by OLAP committee. Firstly, an adequate cube model for MOLAP systems and the corresponding MDX query templates for APB-1 benchmark have been proposed, in which the shortcomings of APB-1 benchmark and their ROLAP star model are discussed during the cube model designing process. Then, experimental results come from new MOLAP model and ROLAP model implementation are compared under the same testing dataset and parameters in order to validate the accuracy of MOLAP model and the design correctness of MDX query templates. At last, according to the purpose of performance comparisons of OLAP systems, the OLAP performance testing procedure and its testing implementation framework for ROLAPMOLAP systems are proposed, in which various modules are explained, and a set of parallel queries has been tested on commercial DBMSs to verify the effectiveness of this framework. The achievement of this study will provide the technical and implemental supports for the performance testing of ROLAPMOLAP products.