‘Green Revolution’ was adopted as National Agricultural Policy by the Government of India to meet the requirements for food grain self-sufficiency. Improved seeds, fertilizers and irrigation were provided as science and technology (S&T) inputs to the soil resource base of the country. Land use changes in order to increase the area of land under cultivation have resulted in sectoral imbalances. The introduction of irrigation, fertilizers and new seed varieties using the district as a unit of agricultural input and management, without consideration of moisture and pedogenetic requirements at the micro-level, have resulted in ecological imbalances and environmental degradation in some parts of the district. Intensive irrigation has resulted in water mining in some blocks, and over a period of years the upper aquifer has been lost and in parts of the Aligarh district water-logging and salinity development has taken place.This paper presents the concept of spatial information systems for sustainable agriculture to eliminate the overdose effect of S&T inputs in a land system such as the fragile ecosystem at the semi-arid border in India.The present study has attempted to integrate the natural resource endowments of lithology, soil, ground-water and geomorphology into terrain mapping units (TMUs). The geological, geomorphological and soil maps were generated using remotely sensed data. For ground-water, well logging was carried out and secondary data regarding depth to water-table were collated. The spatial modelling for depth to water-table was carried out using SURFER software. The files on geology, geomorphology, soil and depth to water-table were digitized and integrated to generate an illuminated model characterized by areas of homogeneity in terms of geology, geomorphology, soil and ground-water conditions to map the TMUs. Intergraph's Modular GIS Environment (MGE) software was used to integrate the multiple data base.For sustainable agriculture the natural boundaries represented by TMUs will be more appropriate as spatial units for the application of agricultural inputs and management practices for the land system, rather than villages, blocks and districts which have cultural boundaries. The methodology proposed will be useful for all ecosystems in semi-arid and arid regions.