Conventional formula fertilization tends to calculate regional rate of fertilizer application by means of analyzing spatial distribution of regional cultivated land productivity combined with experiment data. However, as environment information of cultivated land is incomplete due to limitation of traditional cultivated land management technology and data acquisition, uncertainty of rate of fertilizer makes it hard to define the interval of formula fertilization and support the regional fertilization task. With the technique of spatial analysis and multiscale uncertainty theory, conventional fertilization can be optimized. Four steps are involved to calculate regional formula fertilization interval based on conventional formula fertilization: (i) To simulate cultivated land productivity according to EGLSN Model, and make it crop target field; (ii) To determine rate of fertilizer according to target field to define cultivated land productivity fertilizer interval and mid-value; (iii) To define region fertilizer interval length and value of region varying with scales as environment information becomes complete gradually; (iv) To apply block fertilizer combined with conventional formula by soil testing. Multiscale optimizing formula fertilization system has been established by using the ArcEngine as a platform to integrate the methods, which is applied in Xinjiang County, Shanxi Province, in order to optimize the existing fertilization formula in study area. It showed that the optimized formula fertilization had more spatial details of productivity than the original one. And the new method is available to support formula fertilization in any region or the block with uncertain environment information. It is therefore concluded that the proposed method has the potential for popularity, which provides a multiscale, multiple-factor and standardized formula fertilization method.