ABSTRACT: In national rubber plantation as well as private, one of the activities that has been done annually is estimate production for the following year based on certain block of rubber plantation. Establishing the estimation of the production according to land and plants potential is important because the estimation of production which is not appropriate will cause the production of following year disturbed even the economical life of the plant will be reduced. The modeling estimation of rubber tree production based on satelit, genetic potential and field unit data aimed to (1) examining the ability of remote sensing and geographical information system to identification of rubber tree and manage the production in the rubber plantation (2) examining the relationship the variation of the spectral reflection value (band VNIR), vegetation index, leaf area indexs (LAI) toward rubber plant production (3) making estimation of rubber tree production model, based on satelit data, plant genetic potential (clone variety and physiological quality), and environment conditions (climate and soil). Determining production method are divided into two, those are qualitative and quantitative. Qualitative is done by ranging the crown density based on NDVI with map of land capability. The quantitative models is done based on raster spatially model with the smallest unit is 15 x15 meter. The factor which can be used as a model is the crown density level, spectral reflection value that sourced from Satelit data (ASTER), genetic potential (clone variety and physiological quality), and field unit data (soil and climate). Rubber identification using satelit data based on visual interpretation and NDVI classification showed 75,52 % and 99,90 %. The regression result showed that the highest determination coefficient (r2) is the red wave length with the age of percent of age off rubber tree coefficient about 5-15 years is 79,3% and the age of 6-25 years is 97,6%. The production models with the percentage of highest accuracy is model based the red wave length, genetic potential, and field unit data with percentage accuracy 74,00% ( first model ), 68,73% ( second model) and 64,81% (third model ).