Abstract: Because static materialized views selection algorithm has many shortcomings，such as larger search space，higher time consumption and excluding query probability and distribution，and the changes in data sources can’t be reflected in data warehouse immediately.In view of these，this paper implements dynamic adjustment for static materialized views selection algorithm according to CVLC and IGA，that is，CNUMV algorithm.The algorithm has been proved in reducing search space and time consumption by the experiment.Most of all，because the algorithm considers materialized views mutual relations in influencing view benefit.Consequently，the algorithm can be dynamicly adjusted online and obtains anticipative purpose.