Abstract: In order to help users to quickly retrieve agricultural information, a multidimensional classification model of agricultural information was designed according to the new characteristics of the users’ demand for agricultural information. Then the multidimensional classification model scheme based on OLAP was also put forward from the perspective of information efficient storage and processing. Finally Hunan rape network information was taken as an example to collect experiment data and classify the correlative information to realize the storage structure of agricultural information classification based on OLAP. Experimental results showed that: (1) the improved multidimensional model of agricultural information classification could more effectively express the information collected randomly from agricultural network; (2) the dimensions of industry, time, region and form were objective, and the dimensions of information attribution, industrial chain and information content were subjective; (3) empty value “null” could be used to express the value when the information was not determined or there was no data in the dimension; (4) agricultural science and technology information had no data in time and space dimension. OLAP multidimensional agricultural information classification model designed in this paper could not only view relevant data from each dimension, but also extract data from different dimensions to analyze in depth, which laid an important foundation for personalized agricultural information push service. Realization of multidimensional agricultural information storage structure was necessary for data analysis and dimension optimization in follow-up agricultural information work.