Data is the foundation of scientific research, and in the mean time, it is the result of scientific research. Data sharing can make data more usefully valuable to a great extent. Data sharing is a systematic engineering which includes not only data itself but also software platform for data sharing, technical standards, sharing policy, administrative mechanism and data service. Data is the most basic and important element in data sharing engineering. Geo-science data have attributes of space, time, integration, great capacity and multi-sources. It is costly work to capture geo-science data. Therefore, it is more necessary to share them among different organizations or scientists. Tabular data, remotely sensed data and vector data are most important and often used types of geo-science data. Because of limitations of various kinds of original data, such as statistical data, remotely sensed spectral data(e.g. TM, NOAA-AVHRR,MODIS), map data and so on, besides them, series of data sets should be developed to meet the need of basic and forward research for geo-science. By means of spatialisation technique, observed data from observation stations can be used to calculate data in areas without observation stations and statistical data from administrative divisions can be re-assigned to regular grid, for example, one kilometer by one kilometer. Through various algorithms, remotely sensed spectral data can be used to produce data sets with obvious geo-science sense, for example, Vegetation Index (VI) and Net Primary Production(NPP).The software platform is important technological support for geo-science data sharing. It should be easy to use and include the following fundamental functions: (a) users management; (b) data catalogue query; (c) metadata management; (d) data query and view; and (e) data download.In order to promote scientific data sharing, relevant policies, measures and standards should be drawn up. Reasonable appraising system for people who are engaged in work of data should be set up to encourage they devote themselves to data sharing. 