A well-known challenge in data warehousing is the efficient incremental maintenance of data cube in the presence of source data updates. In this paper, we present a new incremental maintenance algorithm developed from Mumick’s algorithm. Instead of using one auxiliary delta table, We use two to improve efficiency of data update. Moreover, when a materialized view has to be recomputed, we use its smallest ancestral view’s data, while Mumick uses the fact table which is usually much lager than its smallest ancestor. We have implemented this algorithm and found the performance has a significant improvement.