To avoid the subjectivity and difficulty of dealing with high dimensional dataset for conventional methods, a set of evaluation index system of cultivated land intensive use was built. Based on data of land and socioeconomy from 1996 to 2009 in Changsha-Zhuzhou-Xiangtan (Chang-Zhu-Tan) urban agglomeration and in 2009 in each county, quantitative comprehensive evaluation of sequential variation and spatial differentiation was made by a projection pursuit model based on genetic algorithm. The results were obtained as follows. (1) The main factors affecting cultivated land intensive use were multiple-crop index, application quantity of chemical fertilizer, total power of agricultural machinery and plant production for per unit cultivated land area in Chang-Zhu-Tan urban agglomeration. (2) From 1996 to 2009, the degree of cultivated land intensive use in Chang-Zhu-Tan urban agglomeration took the year 2002 as the demarcation point, and presented a phased development trend with "stable-rapid growth". (3) In 2009, spatial differentiation characteristic of the degree of cultivated land intensive use declined gradually from suburbs to middle suburbs and outer suburbs in Chang-Zhu-Tan urban agglomeration; (4) Compared with conventional evaluation methods, the projection pursuit method was of impersonality and credibility, abundance in characterization information, with reasonable results and high validity in evaluation of cultivated land intensive use.