Abstract: Abstract: China has fed about 20% of the world population with only 8% of the world land. Actually, the proportion of arable land is much lower than 8% because most of the land is unavailable for agriculture activity, such as mountains, deserts and so on. Since the reform and opening-up policy was proposed in 1978, China has undergone rapid urbanization, which has become a great threat to food security. In this case, Chinese government proposed a new policy about protection of permanent basic farmland in response to the increasing urban growth. It is important and urgent to zone the permanent farmland for restraining the pell-mell urban expansion and improving the protective efficiency, which is of great significance to food security for China. Whether the basic farmland protection is permanent or not, the core problem is to identify which arable land should be protected. In general, the core elements for identifying permanent basic farmland are the quality and the spatial pattern of arable land, which mainly determines the zoning pattern of permanent basic farmland protection. This aims to contradict the conflicts between farmland protection and urban growth. And there are many objectives should be considered when identifying the permanent basic farmland in actual engineering practice. For example, a contiguous pattern is more preferred for modernizing the agricultural sector. Therefore, it is a complex data mining problem to identify and zone permanent basic farmland from land-use status quo database. In this paper, the result of farmland utilization grade was applied to measure the quality of arable land. Cellular automata was further developed to simulate the conflict between farmland protection and urban growth, and then the arable lands located within the simulated growth area can be easily detected and excluded from protection. However, those arable lands around cities and traffic lines with the high quality may be also allowed for urban growth. This may disobey the new policy of farmland protection enacted by government. Whether the arable lands of high quality around cities and traffic lines are protected or not is greatly determined by local government. In order to zone the plausible protection pattern, a seed search algorithm was improved and induced. A series of factors including the quality of arable land, coordinated pattern of urban growth, landscape connectivity of zoned protection area, and topographic constraints were further incorporated to derive the relatively best protection pattern. Moreover, artificial neural network algorithm was used to forecast the protection stress of basic farmland due to the increasing urban expansion. An intelligent zoning tool (iZone) was developed using the component technology of ArcGIS and C#. This tool was used to identify universe basic farmlands with the integration of the above mentioned data mining models. Jinli town of Gaoyao district located in Guangdong province was further selected as a case study area to test the iZone's performance. The quantitative comparison between the pattern identified by expert's work and that obtained by iZone was also carried out. Results demonstrated that iZone can retrieve a better protection pattern of permanent basic farmland from land-use status quo database, which can efficiently avoid the subjectivity of artificial work. iZone can identify the permanent basic farmland under the support of complex geographic computation technologies. It is of some practicability in decision-making for permanent basic farmland protection.