Remote sensing technology is increasingly used in mapping and monitoring ecosystems as well as providing input datasets for analysis of ecosystem functions or dynamics. As a source of information on ecosystem classification and evaluation, remote sensor data is particularly important at the scale of the landscape, where has few field data. However, each type of remote sensing image often developed an independent classification system to satisfy the needs of the majority of users, guidelines of criteria for certain utility. Although there are many similarities exist in different systems, it is hardly be comparable or interchangeable in general cases. There is, in fact, no logical reason to expect that one detailed ecosystem inventory should be adequate for multi-sources remote sensing data. But an ecosystem categorization system that takes whole consideration of remote sensing information, ecological factors and type or degree of human intervention is able to improve the efficiency of ecosystem analysis using different satellite imagery. Aim at this objective, this paper compiled a ecosystem classification system in China based on medium-resolution remote sensor data. The system includes 9 types, including forest, shrubland, grassland, wetland, cropland, urban land, desert, glacier/ firn, bare land, in the first level categories, 21 subordinate types in the second level categories and 46 types in the third level categories. On the base of 38 third level categories from ChinaCover2010 project, we reclassified this categories in accordance with characteristics of ecosystem types and subdivided the third level categories into 46 by add up to many supplemental information such as climate factors and site condition of vegetations. To interpret the usage and the effectiveness of the system, ecosystem configurations in three regions, Hainan island, Inner Mongolia and Gansu province, have been calculated. Results shown ability to depict zonal features of ecosystem and to use in further ecosyste m analysis. It still indicate that the discrepancy of spatial resolution occurs between ecological factor data and remote sensor data and more detailed levels have not been established which we would take deeper researches in certain areas.