Rapid industrialization and urbanization have brought a series of environmental problems related to soil pollution in the Pearl River Delta region. A systematic and synthetic evaluation of the cultivated land quality is of great significance to guide the rational utilization and protection of cultivated land, as well as to realize the comprehensive balance and management of cultivated land. However, the soil pollution factors have seldom been considered in cultivated land quality evaluation system, which was also short of a sensitivity analysis of the index weight. Based on GIS (geographic information system) multi-objective decision analysis, the paper conducted a study on the evaluation of cultivated land quality and the sensitivity analysis of index weight by taking Zengcheng District of Pearl River Delta as an example. With soil fertility factor and environmental assessment index taken into consideration, the comprehensive evaluation index system of cultivated land quality was constructed from 4 aspects: Soil physical and chemical properties, agricultural production conditions, location conditions and soil environment, so as to analyze the general characteristics of cultivated land quality and the spatial layout rule in Zengcheng District, and the weight sensitivity analysis of OAT (one-at-a-time) method was adopted to evaluate the influence degree of each index weight's uncertainty on the evaluation results. In order to reflect the trend and regularity of the changon cultivated land quality and its spatial pattern, the weight value of only one factor changed at a time while the other factors remained unchanged. When the value of RPC (range of percent change) and IPC (increment of percent change) were at ±30% and ±2% respectively, and every criterion factor was applied to all the evaluation indicators as the main change factor, a total of 420 groups of weight values were generated, among which every group of weights were calculated to produce a new evaluation result of the comprehensive quality of cultivated land. The main results of the study were as follows: 1) The cultivated land in Zengcheng District had a good overall quality, but high quality cultivated land was lacked. The grades of cultivated land quality were mainly concentrated at the 2nd and 3rd grade, which accounted for 30.88% and 31.69%, respectively. The proportion of middle and low yield field was relatively large. 2) From the perspective of the spatial distribution, the cultivated land with better quality was mainly distributed in the southern plains, the basins with better irrigation conditions like Zengjiang River basin, East River basin and Xifu River basin, as well as North River valley. The comprehensive quality of most cultivated land in Zhengguo, Xiaolou and Paitan was not high, and the overall quality of cultivated land in Xintang Town was poor. 3) The changes of weights had a certain influence on the quality distribution of cultivated land. Seen from the spatial distribution of change rate of cultivated land quality generated from different weight distribution, the spatial differences of the calculated results were rather big. When the weight increased or decreased by the same value, the same factor showed the same sensitivity to the quality evaluation results of cultivated land. 4) The maximum MACR (mean absolute change rate) value of 3.558 2% was much lower than the corresponding weight change rate of 30%, showing that the evaluation results were relatively stable, and the quality of cultivated land in Zengcheng District was overall stable. In conclusion, in the comprehensive evaluation of cultivated land quality, the soil environmental factors should be taken into consideration, and a quantitative analysis of micro-data of the soil pollution situation should be conducted to make the evaluation system more comprehensive and more scientific. The sensitivity analysis of the index weight can verify the reliability of the evaluation results, which can help the relevant departments make better decisions, and reduce the uncertainty in the spatial multi-criterion decision-making. [ABSTRACT FROM AUTHOR]