Abstract: Land reallocation in land consolidation is one of the most effective instruments to ameliorate farmland fragmentation. However, the unwillingness of farmers coupled with the inefficiency of reallocation method makes a great obstacle for promotion of it. Therefore, involvement of farmers’ preferences and introduction of new technology with higher efficiency are necessary to get out of the plight. Taking farmland consolidation project in Huangling, Pengze county of Jiangxi province as a study area, in this paper, we constructed a new land reallocation model using transportation model of linear programming aiming at two main preferences of farmers in project area: to minimize the movement of land parcels and to minimize the distance between settlements and land parcels. Having determined the situated land blocks of land parcels, a land partitioning algorithm based on Python was proposed to realize the high efficiency and accuracy of the specific position, the area and shape determination of parcels. Also, a systematic comparison between this new model and the traditional one based on drawing lots was conducted in this paper from three aspects: 1) the amelioration of farmland fragmentation; 2) the change of parcels’ location, and 3) the efficiency and accuracy of land reallocation. The results showed that both models can significantly reduce the number of land parcels, thus relieving land fragmentation. New model was slightly better than traditional one in terms of land parcels reducing while for those who still contracted more than one parcels after land reallocation, the spatial distribution resulted from new models was more scattered. In addition, the new model could reduce the average distance between land parcels and settlements in study area from 479.71 to 445.39 m, while it rose from 479.71 to 556.04 m in traditional model. For the change of location of land parcels, after land reallocation, 75.04% of land parcels in new model remained in the same land block prior to land reallocation, while only 31.06% of them remained in the same land block in the traditional model, attesting that linear programming guaranteed a high degree of meeting farmers’ preferences. New model was more accurate compared to the traditional model. As for the efficiency, two models prevailed in different stages of land reallocation respectively. Traditional model was more efficient in land redistribution due to briefness of the procedure, while new model had significant high efficiency in land partitioning stage. It was concluded that the new model applying linear programming can well meet the needs of farmers, facilitating the promotion of land reallocation, especially in small scale land consolidation project. The introduction of GIS and computer programming also guaranteed the high accuracy and efficiency when compared to the way based on the experience and knowledge of experts. However, traditional model still had its applicable area, especially in places lacking of technical power and highly respected procedural justice. An abundant method system of land reallocation can make sure that the application of a specific model was based on the specific local condition, therefore, the development of new method was one of the key to promote large scale land reallocation in land consolidation.