A SVM (Support Vector Machine)-based method for productivity evaluation of Standard Cultivated Land (SCL) and a GASA-optimized algorithm for selecting of SVM parameters is put forward in this paper. Based on determination of the indices for productivity evaluation and grading of SCL, this method first made use of the data of the samples in the farmland productivity survey and its evaluation results of traditional integrated productivity factors method in building up a SVM sample set, trained SVM with the GASA-optimized algorithm, and set up a SVM model for evaluation and grading of SCL. This method was tested on productivity evaluation of SCL in Lucheng District of Wenzhou City, Zhejiang. The results indicate that the second grade and third grade of SCL accounts for 45.04% and 54.96%, respectively, of the total SCL in area ( 115.7 hm² ) which the test samples stand for. The evaluation using this method reached 100% in accuracy. The evaluation of the test samples using the BP networks method was only 90% in accuracy. The findings show that the SVM method is much higher in accuracy than the BP networks method, and therefore this effective new method can be used efficiently to evaluate and grade SCL in productivity．