Abstract: Territorial planning needs overall consideration of various kinds of territorial resources and an integrated application of land, water, minerals, industrial layout, population distribution and economic subject data. The data are scattered in different departments but have some common characteristics: spatial data constitute the data subject, and the attribute data are closely related to spatial data, decentralized collection, space partition, and theme layers. In order to solve the problem of diversity in data format, production, storage and usage, the authors, on the basis of analyzing the characteristics of GML, proposed a GML-based territorial planning multi-source heterogeneous data integration model, analyzed the process for multi-source heterogeneous data integration, and achieved the conversion of the MapGIS data into GML data. Tests prove that the method can to some extent meet the demand of territorial planning application.