A Database grid can provide a robust distributed environment for accessing and manipulating different database resources. While current challenges focus on managing database resources and ensuring the quality of data integration. This paper describes a Service Level Agreement (SLA) based data integration model, which includes SLA-based Resource Description Model, SLA-based Query Model, SLA-based Matching Algorithm and SLA-based Evaluation Model. The SLAs of a database resource is defined by SLA-based Resource Description Model. The SLA-based Query Model is defined to normalize the various types of queries. The SLA-based Matching Algorithm has been proposed to match schemas of resources for each subquery, and then the quality of database resources is evaluated by SLA-based Evaluation Model to obtain the final query solutions. Experimental results have shown that SLA-based Data Integration Model can provide more stable database resources and save time cost in data integration applications as well.