Data extraction and transform are based on the mapping relation between the new system database and old system database. In generally, transform also includes data cleaning, which is to clear the dirty data that is from original database, Before data cleaning, analysis of data quality is demanded. As long as the dirty data is found, it can be cleared. Data loading is to load the data which has been extracted and transformed into the destination database through loading tools and SQL sentences programmed manually. This paper detailedly analyzes the model of data integration in the bank credit system , in which the process and methods of data extraction,conversion and load have been studied deeply, and the steps of data integration also have been presented.